CN112464712A - Rotating machine fault diagnosis method based on blind extraction algorithm - Google Patents

Rotating machine fault diagnosis method based on blind extraction algorithm Download PDF

Info

Publication number
CN112464712A
CN112464712A CN202011128797.5A CN202011128797A CN112464712A CN 112464712 A CN112464712 A CN 112464712A CN 202011128797 A CN202011128797 A CN 202011128797A CN 112464712 A CN112464712 A CN 112464712A
Authority
CN
China
Prior art keywords
signal
signals
separation
matrix
fault
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
CN202011128797.5A
Other languages
Chinese (zh)
Other versions
CN112464712B (en
Inventor
汪抑非
李创
汤中彩
龚亦昕
刘唐丁
王绪康
柴秋子
沈新荣
杨春节
黄志龙
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.)
Hangzhou Zeta Energy Saving Technology Co ltd
Zhejiang University ZJU
Original Assignee
Hangzhou Zeta Energy Saving Technology Co ltd
Zhejiang University ZJU
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 Hangzhou Zeta Energy Saving Technology Co ltd, Zhejiang University ZJU filed Critical Hangzhou Zeta Energy Saving Technology Co ltd
Priority to CN202011128797.5A priority Critical patent/CN112464712B/en
Publication of CN112464712A publication Critical patent/CN112464712A/en
Application granted granted Critical
Publication of CN112464712B publication Critical patent/CN112464712B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G01M13/04—Bearings
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G01M13/04—Bearings
    • G01M13/045—Acoustic or vibration analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Signal Processing (AREA)
  • Acoustics & Sound (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)

Abstract

The invention relates to a mechanical fault diagnosis technology, and aims to provide a rotary mechanical fault diagnosis method based on a blind extraction algorithm. The method comprises the following steps: multi-channel vibration measurement of rotating machinery; setting a reference signal based on fault information; extracting the separation signal by using a blind extraction algorithm; calculating the energy ratio of the separation signal; and (4) training and diagnosing a rotary machine support vector machine diagnosis model. The method comprises a series of steps from signal acquisition to fault diagnosis and the like, solves the problem that the conventional blind source separation algorithm is only applied to artificial diagnosis and is rarely applied to the aspect of artificial intelligent diagnosis, and improves the accuracy of fault diagnosis of the rotary machine to a greater extent. Therefore, compared with the fault diagnosis algorithm of the rotating machinery adopted in the prior art, the fault diagnosis method has better practical value.

Description

Rotating machine fault diagnosis method based on blind extraction algorithm
Technical Field
The invention relates to a mechanical fault diagnosis technology, in particular to application of signal processing and artificial intelligence to mechanical fault diagnosis, and particularly relates to a rotary mechanical fault diagnosis method based on a blind extraction algorithm.
Background
Rotary machines are a very important and widely used class of mechanical devices in industrial fields, such as fans, water pumps, compressors, etc., and play an important role in various industrial fields. During the working process of the rotary machine, larger or smaller faults are inevitably generated, which cause economic loss and casualties, so that the fault diagnosis of the rotary machine is very important in the industrial field, and the research on the fault diagnosis method of the rotary machine is never stopped.
At present, fault diagnosis for rotary machines is mostly based on vibration signals, when the rotary machines are subjected to vibration testing, the measured signals are signals obtained by mixing a plurality of vibration sources in the rotary machines, and the difficulty of fault feature extraction and fault diagnosis is undoubtedly increased. With the development of artificial intelligence, a plurality of rotary machine fault diagnosis algorithms based on artificial intelligence appear, and compared with the traditional artificial diagnosis, the artificial intelligence fault diagnosis method has the advantages of high accuracy, good timeliness and the like.
The blind source separation algorithm is based on the independence assumption of the vibration source, and under the condition of no priori knowledge, the separation signals with the same generalized form as the vibration source signals are separated, so that the fault characteristics are more obvious. However, the blind source separation algorithm has uncertainty of amplitude and sequence, and is difficult to be used as an input vector of the artificial intelligence diagnostic algorithm. Therefore, an artificial intelligence diagnostic algorithm for rotary machines based on a blind source separation algorithm cannot be achieved in the industry at present. Most of the current fault diagnosis algorithms based on blind source separation are limited to artificial fault diagnosis, and have the defects of poor accuracy, low efficiency and the like. The invention designs an artificial intelligence fault diagnosis system based on a blind extraction algorithm for rotating machinery, which combines the prior knowledge and the traditional blind source separation technology to successively extract separation signals with the same generalized form as the specified vibration source signals and calculate the energy ratio of the separation signals, and combines the blind extraction algorithm and a support vector machine method to greatly improve the accuracy of fault diagnosis.
Disclosure of Invention
The invention aims to solve the technical problem that blind source separation cannot be combined with an artificial intelligence algorithm to carry out fault diagnosis on rotating mechanical equipment, overcomes the defects in the prior art, and provides a rotating mechanical fault diagnosis method combining blind extraction and the artificial intelligence diagnosis algorithm.
In order to solve the technical problem, the solution of the invention is as follows:
the rotary machine fault diagnosis method based on the blind extraction algorithm comprises the following steps:
(1) multi-channel vibration measurement of rotating machinery
Arranging N vibration sensors on a bearing seat on a rotary machine, wherein N is more than 1;
(2) setting based on fault information reference signal
Setting the number of the separation signals to be extracted as N (the number of the extracted separation signals is required to be consistent with the number of the sensors in the step (1)) according to the number of the sensors; for eachA separate signal, a reference signal R having the same frequency as the frequency of the separate signal is setiWherein i is 1, 2.. times.n, the reference signal RiThe method provides prior knowledge for a blind extraction algorithm, and generally adopts a square wave form;
(3) extraction of separated signals using blind extraction algorithm
The blind extraction algorithm is improved on the basis of the traditional blind source separation algorithm, certain priori knowledge is added, the vibration source signals are extracted successively according to the priori knowledge, and the sequence uncertainty in the traditional blind source separation is avoided, so that the blind extraction algorithm can be combined with an artificial intelligence algorithm to carry out fault diagnosis on rotary mechanical equipment.
The observation signals are collectively denoted as X (t) ═ x1(t)，x2(t)，...，xi(t)]T1, 2.. N, wherein xi(T) is an observed signal component, which is a vibration signal measured by the ith vibration sensor, and T represents the transpose of the matrix (the same applies hereinafter);
suppose that the vibration source signal of the rotary machine is s (t) ═ s1(t)，s2(t)，...，si(t)]T1, 2, N, wherein si(t) represents the ith vibration source signal, and the separation signal is y (t) ═ y1(t)，y2(t)，...，yi(t)]T1, 2, N, wherein yi(t) represents the ith split signal, then:
X(t)＝A×S(t) (1)
wherein A is a mixing matrix;
by separating the signal components yi(t) statistical independence between and added a priori knowledge as optimization criteria; the priori knowledge is knowledge of fault signal frequency and the like of the rotating machine, and is added in the form of the reference signal in the step (2); then, through a gradient descent algorithm, finding a separation matrix W which maximizes the optimization criterion, so that the separation signal Y (t) and the vibration source signal S (t) have the same generalized form; the calculation formula is as follows:
Y(t)＝W×X(t)＝W×A×S(t) (2)
(4) energy ratio calculation of separated signals
Each separation signal y is calculated by equation (6)i(t) the energy ratio E in the observed signal X (t)i；
Figure BDA0002734381200000021
In the formula (6), | · non-woven phosphor2Represents a 2-norm; x-i＝X-<X，wi> w i1, 2.. N, for the observation signal X minus the separation signal YiThe contribution in the X-ray spectrum of the light,<·>representing an inner product operation;
the energy of each separated signal is compared with EiThe sum being a matrix form E ═ E1，E2，...，Ei]T1, 2, N, using the energy ratio matrix E as an input feature vector of the support vector machine for diagnosis;
(5) rotary machine support vector machine diagnostic model training and diagnosis
At present, a plurality of fault diagnosis models are used as neural network models, and the neural network models are widely applied due to good robustness, but the neural network needs a large amount of fault data as support, but in practice, the neural network does not have enough fault data, and the support vector machine can achieve good accuracy under the condition of less fault data.
Adopting a directed acyclic graph DAG support vector machine to realize multiple classifiers, and calling k-1 two classifiers to classify the faults if the number of the fault types of the equipment is k;
converting the multiple classifiers into a solution quadratic optimization problem, and solving the problem in the following way:
Figure BDA0002734381200000031
wherein m is the number of samples, αi，αjI, j groups of samples x respectivelyi，xjCorresponding Lagrange multiplier, yi，yjThe fault type values corresponding to the ith and jth groups of samples respectively; k (x)i，xj) Is a kernel function;
obtaining an optimal Lagrange multiplication subset through solving, and completing the establishment and training of a model; after the training is completed, the model is used for diagnosing mechanical faults in the operation process of the equipment.
In the invention, in the step (1), in order to ensure that the measured signals include all vibration source signals of the rotary machine, the N sensors should be respectively installed on different bearing seats as much as possible.
In the present invention, in the step (3), a blind extraction algorithm based on the measure of correlation between negative entropy and convolution is adopted, and the specific steps include:
(3.1) based on the observed signals x measured by the plurality of vibration sensorsi(t), forming an observed signal matrix:
X(t)＝[x1(t)，x2(t)，...，xi(t)]T，i＝1，2，...，N；
(3.2) establishing a spheroidizing matrix B corresponding to the observed signals to eliminate each observed signal component xi(t) correlation between the signals, and noting the observed signal matrix after spheroidizing as
Figure BDA0002734381200000032
Then:
Figure BDA0002734381200000033
(3.3) in the blind extraction algorithm, an objective function needs to be set as an optimization criterion, and the purpose is to enable statistics to have independence and formulate prior knowledge so as to facilitate calculation; compared with the traditional blind source separation algorithm, the target function in the method is added with the prior knowledge item, which is the key for combining the blind extraction algorithm with the artificial intelligent diagnosis algorithm.
The specifically set objective function is:
Figure BDA0002734381200000034
in the formula (I), the compound is shown in the specification,
Figure BDA0002734381200000035
representing a distance measure between two signals as a priori knowledge term; j (-) represents negative entropy; w is aiA separation matrix component corresponding to the ith reference signal; a and b are positive real numbers respectively representing
Figure BDA0002734381200000041
And
Figure BDA0002734381200000042
the size of the proportion occupied in the objective function; τ denotes the time delay, τ ∈ [0, t ∈ >]；
(3.4) the objective function in the formula (4) is optimized by the Newton iteration method by finding the proper separation matrix component wiLet the objective function F (w)i) Maximizing, after iteration is completed, obtaining all optimized separation matrix components wiThe sum being a separation matrix W ═ W1，w2，...，wi]T，i＝1，2，...，N；
(3.5) use formula (5)
Figure BDA0002734381200000043
Obtaining a final separation signal Y (t) ═ y1(t)，y2(t)，...，yi(t)]T，i＝1，2，...，N。
In the present invention, in the step (5), when the multiple classifiers are implemented by using a directed acyclic graph DAG support vector machine: importing data into a root node, dividing the input data into two subclasses by a support vector machine model in the root node, and further dividing the two subclasses by using the support vector machine models on the child nodes; this loops until only one category is included in the subclass.
Compared with the prior art, the invention has the beneficial effects that:
the rotary machine fault diagnosis method based on the blind extraction algorithm comprises a series of steps from signal acquisition to fault diagnosis and the like. The system separates the mechanical fault vibration source signals one by one based on a blind extraction algorithm, and calculates the energy ratio of each vibration source signal in an observation signal, so that the separated fault vibration source signals can be used as feature vectors to be input into a support vector machine model. Therefore, the problem that the conventional blind source separation algorithm is only applied to artificial diagnosis and is rarely applied to the aspect of artificial intelligent diagnosis is solved, and the accuracy of fault diagnosis of the rotary machine is improved to a greater extent. Therefore, compared with the fault diagnosis algorithm of the rotating machinery adopted in the prior art, the fault diagnosis method has better practical value.
Drawings
Fig. 1 is a detailed flowchart of the mechanical failure diagnosis system for a rotary machine according to the present invention.
Fig. 2 shows the mounting of the vibration sensor according to the invention.
Fig. 3 is an observed signal before blind extraction.
Fig. 4 shows the reference signal and the separated signal after blind extraction.
Fig. 5 shows the energy ratio of each separated signal after extraction.
FIG. 6 is a diagram of a multi-class support vector machine.
Reference numerals in fig. 2: a bearing seat 1; the sensor 2 is vibrated.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention provides a rotary machine fault diagnosis method based on a blind extraction algorithm, which comprises the following steps:
(1) multi-channel vibration measurement of rotating machinery
In the embodiment, three vibration sensors are fixedly installed on bearing seats of the rotary machine, and each bearing seat is provided with one vibration acceleration sensor in a vertical upward direction.
The vibration signals acquired by the vibration sensor can be transmitted in a wired or wireless mode, and then are converted to be used as original data of a rotary machine fault diagnosis method. The specific installation position of the vibration sensor is shown in fig. 2, and three collected observation signal components are recorded as x respectively1(t)，x2(t)，x3(t), specific waveforms are shown in FIG. 3.
(2) Setting based on fault reference signal
The setting of the reference signal requires adding the characteristic frequency of the vibration source signal. In this example, the vibration source signals are a bearing outer ring fault signal, a misalignment signal and an unbalance signal. Square wave signals with the same frequency as the vibration source fault signals are adopted as reference signals, the duty ratio is 50 percent, and the duty ratios are respectively marked as R1，R2，R3(ii) a The waveforms of the reference signals are shown as reference signal 1, reference signal 2, and reference signal 3 in fig. 4.
(3) Extraction of separated signals using blind extraction algorithm
The total of the observation signal components collected in step (1) is x (t) ═ x1(t)，x2(t)，x3(t)]TThe signals are formed by mixing vibration source signals inside the rotary machine. In the blind extraction algorithm, the vibration source signals are assumed to be statistically independent of each other and combined with the reference signal R1，R2，R3The separate signals having the same generalized form as the designated vibration source signal are successively decimated. Assuming that the vibration source signal of the rotary machine is S (t) and the separation signal is Y (t), the method comprises
X(t)＝A×S(t) (1)
Wherein A is a mixing matrix.
Blind extraction algorithm with separate signal components yi(t) statistical independence between (t) and added a priori knowledge, which can be knowledge of fault signal frequency of the rotating machine, etc., as optimization criteria. Adding the reference signals in the form of the reference signals in the step (2), and finding a separation matrix which maximizes the optimization criterion through a gradient descent algorithmW such that the separation signal y (t) has the same generalized form as the vibration source signal s (t); the calculation formula is as follows:
Y(t)＝W×X(t)＝W×A×S(t) (2)
according to different quantization standards and prior knowledge addition modes of independence, a plurality of different algorithms exist in the blind extraction algorithm. In this embodiment, a blind extraction algorithm based on the negative entropy and the convolution correlation measure is adopted, and the algorithm steps are described in the following simple example:
step 1: spheroidizing the mixed signals, calculating a proper spheroidizing matrix B, and eliminating the correlation of the signals, namely meeting the requirement
Figure BDA0002734381200000061
In the formula (I), the compound is shown in the specification,
Figure BDA0002734381200000062
indicating a sphering signal after sphering, with no correlation of its components.
Step 2: constructing an optimization equation, and calculating a separation matrix W:
Figure BDA0002734381200000063
in the formula (I), the compound is shown in the specification,
Figure BDA0002734381200000064
is a priori knowledge term and represents a distance measure of the separation signal and the corresponding reference signal, tau represents time delay and tau belongs to 0 and t]。
The objective function in the formula (4) is optimized by adopting a Newton iteration method, and a proper separation matrix component w is searchediLet the objective function F (w)i) Maximizing, i is 1, 2, 3, and after completing iteration, obtaining all optimized separation matrix components wiThe sum being a separation matrix W ═ W1，w2，w3]T。
And step 3: by calculating the score in step 2The separation matrix to calculate the corresponding separation signal: y (t) ═ y1(t)，y2(t)，y3(t)]TThe waveforms of the split signals are shown as split signal 1, split signal 2, and split signal 3 in fig. 4.
Figure BDA0002734381200000065
(4) Energy ratio calculation of separated signals
Since the separation signal y (t) in step (3) has unit variance, the amplitude is not a true magnitude. To ensure that the split signals can be used as input quantities of the support vector machine, the energy ratio E of each split signal is calculated by the formula (6)i。
Figure BDA0002734381200000066
Wherein the content of the first and second substances,<·>represents the inner product; i | · | purple wind2The expression is given in the 2-norm,
Figure BDA0002734381200000067
representing the observation signal X minus the separation signal YiContribution in X.
(5) Training and diagnosing rotary machine support vector machine diagnostic model
The invention adopts a directed acyclic graph DAG support vector machine and a tree form multi-classification method to realize multi-classifiers, namely k-1 two-class classifiers are called to classify fault types, and a k-finger fault type quantity multi-classification basic structure is shown in FIG. 6;
and importing the data into a root node, dividing the input data into two subclasses by a support vector machine model in the root node, further dividing the two subclasses by using the support vector machine models on the child nodes, and circulating the steps until only one class is contained in the subclasses. The method has the advantages that the number of the support vector machines needing to be trained and the number of training samples of each support vector machine are small, all support vector machine classifiers do not need to be traversed during classification, the method has high training speed and classification speed, and the method has more obvious advantages for the classification problem with large number of classes.
Converting the multiple classifiers into a problem of solving quadratic optimization, and solving the problem in the following way:
Figure BDA0002734381200000068
wherein m is the number of samples, αi，αjI, j groups of samples x respectivelyi，xjCorresponding Lagrange multiplier, yi，yjThe fault type values corresponding to the ith and jth groups of samples respectively; k (x)i，xj) Is a kernel function;
in this example, the kernel function is chosen to be
Figure BDA0002734381200000071
Where σ is the bandwidth of the gaussian kernel.
And selecting data under the three fault conditions of bearing outer ring fault, misalignment fault and unbalance fault as training data.
And (5) solving the formula (7) to obtain an optimal Lagrange multiplication subset, and finishing the training of the model.
And (5) inputting the feature vector E in the step (4) into the trained support vector machine model, and outputting a diagnosis result.

Claims (5)

1. A rotating machinery fault diagnosis method based on a blind extraction algorithm is characterized by comprising the following steps:
(1) multi-channel vibration measurement of rotating machinery
Arranging N vibration sensors on a bearing seat on a rotary machine, wherein N is more than 1;
(2) setting based on fault information reference signal
Setting the number of the separation signals to be extracted as N according to the number of the sensors; for each of the split signals, a reference signal R is set having the same frequency as the frequency of the split signaliWherein i is 1, 2,.., N; the reference signal RiThe method adopts a square wave form and is used for providing priori knowledge for a blind extraction algorithm;
(3) extraction of separated signals using blind extraction algorithm
The observation signals are collectively denoted as X (t) ═ x1(t)，x2(t)，...，xi(t)]T1, 2, ·, N; wherein xi(t) is an observed signal component representing a vibration signal measured by the ith vibration sensor; t represents the transpose of the matrix;
suppose that the vibration source signal of the rotary machine is s (t) ═ s1(t)，s2(t)，...，si(t)]T1, 2, N, wherein si(t) represents the ith vibration source signal; the separation signal is Y (t) ═ y1(t)，y2(t)，...，yi(t)]T1, 2, N, wherein yi(t) represents the ith split signal; then:
X(t)＝A×S(t) (1)
wherein A is a mixing matrix;
by separating the signal components yi(t) using statistical independence between the signals and the added prior knowledge as optimization criteria, and finding a separation matrix W which maximizes the optimization criteria through a gradient descent algorithm so that the separation signal Y (t) and the vibration source signal S (t) have the same generalized form; the calculation formula is as follows:
Y(t)＝W×X(t)＝W×A×S(t) (2)
(4) energy ratio calculation of separated signals
Each of the separated signal components y is calculated by equation (6)i(t) the energy ratio E in the observed signal X (t)i；
Figure FDA0002734381190000011
In the formula (6), | · non-woven phosphor2Represents a 2-norm; x-i＝X-<X，wi>wi1, 2.. N, for the observation signal X minus the separation signal YiContribution in X，<·>Representing an inner product operation;
the energy of each separated signal is compared with EiThe sum being a matrix form E ═ E1，E2，...，Ei]T1, 2, ·, N; taking the energy ratio matrix E as an input feature vector of a support vector machine for diagnosis;
(5) training and diagnosing rotary machine support vector machine diagnostic model
Adopting a directed acyclic graph DAG support vector machine to realize multiple classifiers, and calling k-1 two classifiers to classify the faults if the number of the fault types of the equipment is k;
converting the multiple classifiers into a solution quadratic optimization problem, and solving the problem in the following way:
Figure FDA0002734381190000021
in the formula (7), m is the number of samples, αi，αjI, j groups of samples x respectivelyi，xjA corresponding Lagrangian multiplier; y isi，yjThe fault type values, k (x), corresponding to the ith and jth groups of samplesi，xj) Is a kernel function;
obtaining an optimal Lagrange multiplication subset through solving, and completing the establishment and training of a model; after the training is completed, the model is used for diagnosing mechanical faults in the operation process of the equipment.
2. The method according to claim 1, wherein in the step (1), in order to ensure that the observation signal includes all vibration source signals of the rotating machine, the N sensors are installed on different bearing seats respectively; the number of separate signals to be extracted corresponds to the number of sensors.
3. Method according to claim 1, characterized in that said a priori knowledge is a priori gathered fault signal frequencies of the rotating machine collated and referenced with a reference signal RiIs added in the form of (1).
4. The method according to claim 1, wherein the step (3) employs a blind extraction algorithm based on a measure of correlation between negative entropy and convolution, and specifically comprises:
(3.1) observed Signal component x measured from multiple vibration Sensorsi(t), observed signals constituting a matrix form:
X(t)＝[x1(t)，x2(t)，...，xi(t)]T，i＝1，2，...，N；
(3.2) establishing a spheroidizing matrix B corresponding to the observed signals to eliminate each observed signal component xi(t) correlation between; the observed signal matrix after spheroidizing is recorded as
Figure FDA0002734381190000022
Then:
Figure FDA0002734381190000023
(3.3) setting an objective function in a blind extraction algorithm as an optimization criterion, so that statistics have independence and a priori knowledge is formulated to facilitate calculation; the specifically set objective function is:
Figure FDA0002734381190000024
in the formula (I), the compound is shown in the specification,
Figure FDA0002734381190000025
representing a distance measure between two signals as a priori knowledge term; j (-) represents negative entropy; w is aiA separation matrix component corresponding to the ith reference signal; a and b are positive real numbers respectively representing
Figure FDA0002734381190000026
And
Figure FDA0002734381190000027
the size of the proportion occupied in the objective function; τ denotes the time delay, τ ∈ [0, t ∈ >]；
(3.4) the objective function in the formula (4) is optimized by the Newton iteration method by finding the proper separation matrix component wiLet the objective function F (w)i) Maximization; after the iteration is completed, all the optimized separation matrix components w are obtainediThe sum being a separation matrix W ═ W1，w2，...，wi]T，i＝1，2，...，N；
(3.5) use formula (5)
Figure FDA0002734381190000028
Obtaining a final separation signal Y (t) ═ y1(t)，y2(t)，...，yi(t)]T，i＝1，2，...，N。
5. The method of claim 1, wherein in step (5), when implementing the multi-classifier using a directed acyclic graph DAG support vector machine: importing data into a root node, dividing the input data into two subclasses by a support vector machine model in the root node, and further dividing the two subclasses by using the support vector machine models on the child nodes; this loops until only one category is included in the subclass.
CN202011128797.5A 2020-10-20 2020-10-20 Rotating machine fault diagnosis method based on blind extraction algorithm Active CN112464712B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011128797.5A CN112464712B (en) 2020-10-20 2020-10-20 Rotating machine fault diagnosis method based on blind extraction algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011128797.5A CN112464712B (en) 2020-10-20 2020-10-20 Rotating machine fault diagnosis method based on blind extraction algorithm

Publications (2)

Publication Number Publication Date
CN112464712A true CN112464712A (en) 2021-03-09
CN112464712B CN112464712B (en) 2022-07-22

Family

ID=74833980

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011128797.5A Active CN112464712B (en) 2020-10-20 2020-10-20 Rotating machine fault diagnosis method based on blind extraction algorithm

Country Status (1)

Country Link
CN (1) CN112464712B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113743225A (en) * 2021-08-05 2021-12-03 武汉科技大学 VT-ANM-SWT-based rotary machine different vibration source identification method
CN115935153A (en) * 2023-01-06 2023-04-07 香港理工大学深圳研究院 Train bearing fault diagnosis method based on Bayesian blind source separation technology

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140301555A1 (en) * 2011-11-07 2014-10-09 Wayne State University Blind extraction of target signals
CN104390780A (en) * 2014-11-25 2015-03-04 沈阳化工大学 Gear case fault diagnosis method based on blind source separation
CN104655425A (en) * 2015-03-06 2015-05-27 重庆大学 Bearing fault classification diagnosis method based on sparse representation and LDM (large margin distribution machine)
US20150233792A1 (en) * 2012-05-31 2015-08-20 Canrig Drilling Technology Methods and Apparatuses for Defect Diagnosis in a Mechanical System
CN109632312A (en) * 2019-01-22 2019-04-16 北京化工大学 Bearing combined failure diagnostic method based on multiple constraint Algorithms of Non-Negative Matrix Factorization
CN111693311A (en) * 2020-05-30 2020-09-22 杭州哲达科技股份有限公司 Rotary machine fault diagnosis method based on independent component analysis and correlation criterion

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140301555A1 (en) * 2011-11-07 2014-10-09 Wayne State University Blind extraction of target signals
US20150233792A1 (en) * 2012-05-31 2015-08-20 Canrig Drilling Technology Methods and Apparatuses for Defect Diagnosis in a Mechanical System
CN104390780A (en) * 2014-11-25 2015-03-04 沈阳化工大学 Gear case fault diagnosis method based on blind source separation
CN104655425A (en) * 2015-03-06 2015-05-27 重庆大学 Bearing fault classification diagnosis method based on sparse representation and LDM (large margin distribution machine)
CN109632312A (en) * 2019-01-22 2019-04-16 北京化工大学 Bearing combined failure diagnostic method based on multiple constraint Algorithms of Non-Negative Matrix Factorization
CN111693311A (en) * 2020-05-30 2020-09-22 杭州哲达科技股份有限公司 Rotary machine fault diagnosis method based on independent component analysis and correlation criterion

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ZEGUANG YI等: "Mechanical compound faults extraction based on improved frequency domain blind deconvolution algorithm", 《MECHANICAL SYSTEMS AND SIGNAL PROCESSING》 *
汪抑非等: "大型机泵健康监测与诊断***设计与应用", 《工业仪表与自动化装置》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113743225A (en) * 2021-08-05 2021-12-03 武汉科技大学 VT-ANM-SWT-based rotary machine different vibration source identification method
CN115935153A (en) * 2023-01-06 2023-04-07 香港理工大学深圳研究院 Train bearing fault diagnosis method based on Bayesian blind source separation technology

Also Published As

Publication number Publication date
CN112464712B (en) 2022-07-22

Similar Documents

Publication Publication Date Title
US20200271720A1 (en) Method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation
CN111693311B (en) Rotary machine fault diagnosis method based on independent component analysis and correlation criterion
CN112464712B (en) Rotating machine fault diagnosis method based on blind extraction algorithm
CN110334764A (en) Rotating machinery intelligent failure diagnosis method based on integrated depth self-encoding encoder
CN111189638B (en) HMM and QPSO optimization algorithm-based bearing fault degree identification method
CN107680116A (en) A kind of method for monitoring moving object in video sequences
CN109117747A (en) Radar signal classification method based on cyclo-stationary Yu depth convolutional neural networks
CN112257528A (en) Wind power gear box fault diagnosis method based on wavelet transformation and dense connection expansion convolutional neural network
US20220372891A1 (en) Method for stability analysis of combustion chamber of gas turbine engine based on image sequence analysis
CN113887136A (en) Improved GAN and ResNet based electric vehicle motor bearing fault diagnosis method
CN109829494A (en) A kind of clustering ensemble method based on weighting similarity measurement
CN115859142A (en) Small sample rolling bearing fault diagnosis method based on convolution transformer generation countermeasure network
CN112765908A (en) Gas turbine engine combustion chamber stability analysis method based on image sequence analysis
CN111310719B (en) Unknown radiation source individual identification and detection method
CN104809478B (en) A kind of image block method and device towards extensive three-dimensional reconstruction
CN111458145A (en) Cable car rolling bearing fault diagnosis method based on road map characteristics
CN113920348A (en) Wind generating set gear box diagnosis method based on semi-supervised momentum prototype network
CN110929761A (en) Balance method for collecting samples in situation awareness framework of intelligent system security system
CN113758709A (en) Rolling bearing fault diagnosis method and system combining edge calculation and deep learning
CN112541524A (en) BP-Adaboost multi-source information motor fault diagnosis method based on attention mechanism improvement
CN109345537A (en) Based on the SAR image segmentation method that the multiple dimensioned CRF of high-order is semi-supervised
CN114941890A (en) Central air conditioner fault diagnosis method and system based on image and depth blurring
CN114818811A (en) Aircraft engine rolling bearing fault diagnosis method based on twin network metric learning
Wei et al. Application of image recognition based on artificial immune in rotating machinery fault diagnosis
CN116340833B (en) Fault diagnosis method based on countermeasure migration network in improved field

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