CN116859240A - Direct-current brushless motor fault identification method based on multiple sensors and edge calculation - Google Patents

Direct-current brushless motor fault identification method based on multiple sensors and edge calculation Download PDF

Info

Publication number
CN116859240A
CN116859240A CN202310838084.5A CN202310838084A CN116859240A CN 116859240 A CN116859240 A CN 116859240A CN 202310838084 A CN202310838084 A CN 202310838084A CN 116859240 A CN116859240 A CN 116859240A
Authority
CN
China
Prior art keywords
motor
brushless
direct current
neural network
real
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
Application number
CN202310838084.5A
Other languages
Chinese (zh)
Inventor
傅雷
张�浩
柴昊祺
胥芳
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.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
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 Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN202310838084.5A priority Critical patent/CN116859240A/en
Publication of CN116859240A publication Critical patent/CN116859240A/en
Pending legal-status Critical Current

Links

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
    • G01R31/343—Testing dynamo-electric machines in operation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Control Of Motors That Do Not Use Commutators (AREA)

Abstract

The application relates to a direct current brushless motor fault identification method based on multi-sensor and edge calculation, which comprises the steps of demodulating a real-time vibration signal of a direct current brushless motor and acquiring a three-phase current signal of the direct current brushless motor; extracting features according to the real-time vibration signals and the three-phase current signals to obtain motor fault identification features; performing feature fusion on motor fault identification features by using a back propagation neural network to generate a feature fusion result; and (3) reasoning the feature fusion result by using a neural network, and identifying the motor fault state. According to the method, the acquired vibration signals and three-phase current signals are sequentially demodulated, the demodulated vibration signals are obtained simultaneously after the sampling duration of the sensor is finished, the characteristic fusion is carried out by adopting a Back Propagation Neural Network (BPNN), the stability of the characteristics is improved, the characteristics are identified, the fault signal characteristics of the brushless direct current motor can be extracted more quickly, and the fault type can be identified in real time.

Description

Direct-current brushless motor fault identification method based on multiple sensors and edge calculation
Technical Field
The application belongs to the technical field of motor fault identification, and particularly relates to a direct current brushless motor fault identification method based on multi-sensor and edge calculation.
Background
The motor is one of the most important components in the modern industrial system, and the economic loss caused by the failure shutdown of the motor is immeasurable, so that the state-based monitoring and the failure diagnosis are important links for ensuring the normal operation of the motor. The vibration sensor is arranged on the motor to monitor the working condition of the motor, so that the safe operation of the motor is ensured, and the possible fault loss is reduced. The fault diagnosis of the motor comprises three steps of data acquisition and preprocessing, feature extraction and pattern recognition. In the past decades, a great deal of research has been carried out by expert students at home and abroad, and many methods for diagnosing motor faults have been proposed mainly focusing on the extraction and analysis of vibration signals. Such as empirical mode decomposition, wavelet transformation, stochastic resonance, and the like. However, these conventional signal analysis methods require a large amount of a priori knowledge, resulting in low diagnosis efficiency and high workload. In recent years, with the development of artificial intelligence, cloud computing has been applied to fault diagnosis. By collecting big data and sending it to the cloud server, a high performance computer with a large-scale central and graphics processing unit can extract features and evaluate the condition of large-scale machines simultaneously. However, as the number of sensors increases, the bandwidth and power consumed, as well as the storage space and computing resources, of data generated by the cloud server increases significantly and affects the real-time nature of motor fault diagnosis.
Therefore, there is a need for a method for identifying faults of a brushless dc motor, which can identify the operating state of the motor in real time with low occupation of computing resources, and further identify the fault type when the motor fails.
Disclosure of Invention
Based on the above-mentioned drawbacks and deficiencies of the prior art, it is an object of the present application to at least solve the above-mentioned problems of the prior art, in other words, to provide a method for identifying faults of a brushless dc motor based on multi-sensor and edge calculation, which meets the above-mentioned needs.
In order to achieve the aim of the application, the application adopts the following technical scheme:
a direct current brushless motor fault identification method based on multi-sensor and edge calculation comprises the following steps:
s1, demodulating a real-time vibration signal of a direct current brushless motor, and acquiring a three-phase current signal of the direct current brushless motor;
s2, extracting features according to the real-time vibration signals and the three-phase current signals to obtain motor fault identification features;
s3, performing feature fusion on motor fault identification features by using a back propagation neural network to generate a feature fusion result;
and S4, reasoning the feature fusion result by using a neural network, and identifying the motor fault state.
As a preferred embodiment, the demodulation in step S1 is a real-time demodulation, and the real-time vibration signal is subjected to a sequence analysis.
As a preferred embodiment, the feature extraction adjusts the discrete resolution according to the real-time frequency of the dc brushless motor.
As a preferred embodiment, feature extraction extracts features from each phase of the three-phase current signal separately.
As a preferred embodiment, step S2 is preceded by a step S20 of normalizing the real-time vibration signal and the three-phase current signal.
As a preferred embodiment, the back propagation neural network of step S3 is trained using the following method:
collecting vibration signals and three-phase current signals of the direct current brushless motor in various working states;
respectively acquiring a plurality of characteristics from vibration signals and three-phase current signals of the direct-current brushless motor in various working states, and reducing the dimension of the characteristics to 1;
and training the counter-propagating neural network by using a plurality of groups of vibration signals, three-phase current signals and corresponding working states.
As a further preferred embodiment, the neural network determines the operation state of the brushless dc motor according to the feature fusion result of the counter-propagating neural network, and identifies the motor fault state according to the operation state.
Compared with the prior art, the application has the beneficial effects that:
the application provides a brushless direct current motor fault diagnosis method based on multi-sensor and edge calculation, which combines signal frequency domain analysis and neural network lightweight deployment. Sequentially demodulating the acquired vibration signals and the three-phase current signals by adopting a real-time signal demodulation method, and simultaneously obtaining demodulated vibration signals after the sampling duration of the sensor is finished; extracting characteristics of four signal channels from frequency points of a frequency spectrum by utilizing Fast Fourier Transform (FFT), and carrying out normalization processing on vibration signals and three-phase current signals in each motor state to eliminate the influence of signal amplitude; adopting a Back Propagation Neural Network (BPNN) to perform feature fusion, and improving the stability of the features; training and deploying the light neural network to the singlechip to realize motor fault diagnosis, reduce the calculation resource waste caused by data transmission and improve the instantaneity. Compared with other methods, the method provided by the application can extract the fault signal characteristics of the brushless direct current motor more quickly and identify the fault type in real time.
Drawings
FIG. 1 is a flow chart of a method for identifying faults of a brushless DC motor based on multiple sensors and edge calculation;
fig. 2 is a schematic structural diagram of a back propagation neural network according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.
In the following description, various embodiments of the application are provided, and various embodiments may be substituted or combined, so that the application is intended to include all possible combinations of the same and/or different embodiments described. Thus, if one embodiment includes feature A, B, C and another embodiment includes feature B, D, then the present application should also be considered to include embodiments that include one or more of all other possible combinations including A, B, C, D, although such an embodiment may not be explicitly recited in the following.
The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made in the function and arrangement of elements described without departing from the scope of the application. Various examples may omit, replace, or add various procedures or components as appropriate. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Furthermore, features described with respect to some examples may be combined into other examples.
The application provides a direct current brushless motor fault identification method based on multi-sensor and edge calculation, wherein a flow chart is shown in figure 1, and the method comprises the following steps:
s1, demodulating a real-time vibration signal of the direct current brushless motor, and acquiring a three-phase current signal of the direct current brushless motor.
Conventional demodulation uses the hilbert transform, which is a common signal demodulation technique. However, this technique is based on complex convolution operations, requiring considerable memory space and computational resources, and is not suitable for resource-limited embedded systems.
For the above reasons, in the technical solution of this embodiment, step S1 sequentially demodulates the collected vibration signals by using a real-time signal demodulation method. The vibration signal v (t) is composed of a high-frequency component v h (t) and Low frequency component v l (t) modulating, v (t) =v h (t)×v l (t)＝A h cos(ω h t)×[A l cos(ω l t)+B]Wherein A is h And A l For signal amplitude, omega h And omega l For signal frequency, B is assurance A l cos(ω l t)+B>A dc bias of 0.
Step S1 of this embodiment may perform a sequence analysis on a single vibration signal sample using a real-time demodulation method, and the calculation time of each sample may be short, and the calculation may be completed within a sampling interval, and at the end of the sampling duration, the demodulated vibration signal may be obtained simultaneously.
And S2, extracting features according to the real-time vibration signals and the three-phase current signals to obtain motor fault identification features.
Specifically, step S2 includes the following steps:
features of four signal channels are extracted from frequency points of the FFT spectrum. Taking into account the variation of the statistical characteristics with the variation of the motor speed, at a rotation frequency f r The frequency region is partitioned for reference.
This division versus rotational frequency is manifested in the variation of discrete resolution, with the discrete versions of the real-time vibration signal and the three-phase current signal being represented as S n, respectively]，I A [n]，I B [n]，I C [n]The resolution is calculated by the formula Δf=f r and/N, so that the discrete resolution varies with the real-time rotation frequency, and the statistical characteristics are not influenced by the rotation speed of the motor.
For each signal, 10 features are extracted from the spectrum, i.e. 10 features are extracted in each of the three phases of the real-time vibration signal and the three-phase current signal, giving a total of 40 features.
Further, step S20 is further included before step S2, where the vibration signal and the three-phase current signal in each motor state are normalized, so as to eliminate the influence of the signal amplitude.
And S3, performing feature fusion on the motor fault identification features by using a back propagation neural network, and generating a feature fusion result.
The embodiment provides the training method of the back propagation neural network, specifically, the back propagation neural network in step S3 is trained by using the following method:
vibration signals and three-phase current signals under 6 different states of bearing inner ring faults, hall sensor faults, open phases, rotor eccentricity, bearing outer ring faults and normal operation are collected and stored as offline data, and sampling points of each frame of the signals are set to be N=2048.
40 features are extracted from each signal group, fused into an index, and the feature dimension is reduced from 40 to 1. The one-dimensional fusion indicator consists of 6 possible values, corresponding to 6 different motor conditions;
for each motor state, a Back Propagation Neural Network (BPNN) training was performed using 200 sets of signals, the back propagation neural network having a structure schematically shown in fig. 2, and the total number of training samples was 1200. During training, the learning rate was set to 0.1. The number of nodes of the input layer, the hidden layer, and the output layer is i=40, j=10, and k=1, respectively. The target matrix is set as T= [1,2,3,4,5,6] T, which respectively correspond to six states of bearing inner ring fault, hall sensor fault, open phase, rotor eccentricity, bearing outer ring fault and normal operation.
And S4, reasoning the feature fusion result by using a neural network, and identifying the motor fault state. Specifically, in this embodiment, the neural network determines the working state of the dc brushless motor according to the feature fusion result of the counter-propagating neural network, where the working state includes six types of bearing inner ring faults, hall sensor faults, open phases, rotor eccentricity, bearing outer ring faults and normal operation, and if the judging result is not the normal working state, the motor fault state is further identified according to the working state, and a specific fault type is determined.
The application provides a brushless direct current motor fault diagnosis method based on multi-sensor and edge calculation, which combines signal frequency domain analysis and neural network lightweight deployment. Sequentially demodulating the acquired vibration signals and the three-phase current signals by adopting a real-time signal demodulation method, and simultaneously obtaining demodulated vibration signals after the sampling duration of the sensor is finished; extracting characteristics of four signal channels from frequency points of a frequency spectrum by utilizing Fast Fourier Transform (FFT), and carrying out normalization processing on vibration signals and three-phase current signals in each motor state to eliminate the influence of signal amplitude; adopting a Back Propagation Neural Network (BPNN) to perform feature fusion, and improving the stability of the features; training and deploying the light neural network to the singlechip to realize motor fault diagnosis, reduce the calculation resource waste caused by data transmission and improve the instantaneity. Compared with other methods, the method can extract the fault signal characteristics of the brushless direct current motor more quickly and identify the fault type in real time.
The foregoing is merely exemplary embodiments of the present disclosure and is not intended to limit the scope of the present disclosure. That is, equivalent changes and modifications are contemplated by the teachings of this disclosure, which fall within the scope of the present disclosure. Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a scope and spirit of the disclosure being indicated by the claims.

Claims (7)

1. The direct current brushless motor fault identification method based on the multi-sensor and edge calculation is characterized by comprising the following steps of:
s1, demodulating a real-time vibration signal of a direct current brushless motor, and acquiring a three-phase current signal of the direct current brushless motor;
s2, extracting features according to the real-time vibration signals and the three-phase current signals to obtain motor fault identification features;
s3, performing feature fusion on the motor fault identification features by using a back propagation neural network to generate a feature fusion result;
and S4, reasoning the feature fusion result by using a neural network, and identifying a motor fault state.
2. The method for identifying faults of a brushless direct current motor based on multiple sensors and edge calculation as claimed in claim 1, wherein the demodulation in the step S1 is real-time demodulation, and the real-time vibration signal is subjected to sequence analysis.
3. A method for identifying faults in a brushless dc motor based on multiple sensors and edge calculations as claimed in claim 1, wherein the feature extraction adjusts the discrete resolution in dependence on the real time frequency of the brushless dc motor.
4. A method of dc brushless motor fault identification based on multi-sensor and edge calculation as claimed in claim 1, wherein the feature extraction extracts features from each phase of the three-phase current signal separately.
5. The method for identifying faults of a brushless direct current motor based on multiple sensors and edge calculation as claimed in claim 1, wherein the step S2 further comprises a step S20 of normalizing the real-time vibration signal and the three-phase current signal.
6. The method for identifying faults of a brushless dc motor based on multiple sensors and edge calculation as claimed in claim 1, wherein the counter-propagating neural network of step S3 is trained by using the following method:
collecting vibration signals and three-phase current signals of the direct current brushless motor in various working states;
respectively acquiring a plurality of characteristics from vibration signals and three-phase current signals of the direct-current brushless motor in various working states, and reducing the characteristic dimension to 1;
and training the counter propagation neural network by using a plurality of groups of vibration signals, three-phase current signals and corresponding working states.
7. The method for identifying faults of a brushless direct current motor based on multiple sensors and edge calculation as claimed in claim 6, wherein the neural network determines the working state of the brushless direct current motor according to the characteristic fusion result of the back propagation neural network, and identifies the fault state of the motor according to the working state.
CN202310838084.5A 2023-07-10 2023-07-10 Direct-current brushless motor fault identification method based on multiple sensors and edge calculation Pending CN116859240A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310838084.5A CN116859240A (en) 2023-07-10 2023-07-10 Direct-current brushless motor fault identification method based on multiple sensors and edge calculation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310838084.5A CN116859240A (en) 2023-07-10 2023-07-10 Direct-current brushless motor fault identification method based on multiple sensors and edge calculation

Publications (1)

Publication Number Publication Date
CN116859240A true CN116859240A (en) 2023-10-10

Family

ID=88229920

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310838084.5A Pending CN116859240A (en) 2023-07-10 2023-07-10 Direct-current brushless motor fault identification method based on multiple sensors and edge calculation

Country Status (1)

Country Link
CN (1) CN116859240A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117168608A (en) * 2023-11-02 2023-12-05 默拓(江苏)电气驱动技术有限公司 Operation early warning method and system of brushless motor

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117168608A (en) * 2023-11-02 2023-12-05 默拓(江苏)电气驱动技术有限公司 Operation early warning method and system of brushless motor
CN117168608B (en) * 2023-11-02 2024-03-08 默拓(江苏)电气驱动技术有限公司 Operation early warning method and system of brushless motor

Similar Documents

Publication Publication Date Title
Ertunc et al. ANN-and ANFIS-based multi-staged decision algorithm for the detection and diagnosis of bearing faults
Lei et al. EEMD method and WNN for fault diagnosis of locomotive roller bearings
CN108760327B (en) Diagnosis method for rotor fault of aircraft engine
Isham et al. Variational mode decomposition: mode determination method for rotating machinery diagnosis
CN116859240A (en) Direct-current brushless motor fault identification method based on multiple sensors and edge calculation
CN105203346A (en) Fault diagnosis method and system for range hood based on EMD (Empirical Mode Decomposition) noise reduction
CN112857804B (en) Rolling bearing fault diagnosis method, device, medium and computer equipment
CN112633098A (en) Fault diagnosis method and system for rotary machine and storage medium
CN111812507A (en) Motor fault diagnosis method based on graph convolution
Wang et al. Weak fault diagnosis of rolling bearing under variable speed condition using IEWT-based enhanced envelope order spectrum
CN114486263B (en) Noise reduction demodulation method for vibration signal of rolling bearing of rotary machine
CN116595317A (en) Motor fault diagnosis method based on orthogonal stator current GA analysis and SFO-PNN
CN115409052A (en) Fault diagnosis method and system for wind generating set bearing under data imbalance
CN111046790A (en) Pump bearing fault diagnosis method
CN114330430A (en) Elevator fault judgment method and system based on big data characteristic analysis
CN116358873B (en) Rolling bearing fault diagnosis method, device, equipment and medium
CN116383739A (en) Intelligent fault diagnosis method based on domain self-adaption multi-mode data fusion
CN116086800A (en) Composite fault diagnosis method for rolling bearing
CN115931318A (en) Fault intelligent diagnosis method, device, equipment and storage medium
CN116127297A (en) Electromechanical device fault diagnosis method and system
CN114077850B (en) Method for monitoring state of rotary mechanical equipment based on graph data under variable working conditions
Tanaka et al. Application of Kohonen's self-organizing network to the diagnosis system for rotating machinery
Senanayaka et al. Autoencoders and data fusion based hybrid health indicator for detecting bearing and stator winding faults in electric motors
Shen et al. Fault detection method based on multi-scale convolutional neural network for wind turbine gearbox
CN205449521U (en) Range hood failure diagnosis's system based on EMD falls and makes an uproar

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