CN110399986A - A kind of generation method of pumping plant unit fault diagnosis system - Google Patents
A kind of generation method of pumping plant unit fault diagnosis system Download PDFInfo
- Publication number
- CN110399986A CN110399986A CN201910550323.0A CN201910550323A CN110399986A CN 110399986 A CN110399986 A CN 110399986A CN 201910550323 A CN201910550323 A CN 201910550323A CN 110399986 A CN110399986 A CN 110399986A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- optimal
- pumping plant
- combination
- plant unit
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000005086 pumping Methods 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000003745 diagnosis Methods 0.000 title claims abstract description 23
- 238000010801 machine learning Methods 0.000 claims abstract description 32
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 230000002068 genetic effect Effects 0.000 claims abstract description 15
- 238000000605 extraction Methods 0.000 claims abstract description 10
- 239000000284 extract Substances 0.000 claims description 11
- 238000004064 recycling Methods 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000012549 training Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000005259 measurement Methods 0.000 description 7
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 6
- 230000001133 acceleration Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000004888 barrier function Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000003973 irrigation Methods 0.000 description 1
- 230000002262 irrigation Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- 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
- G06F18/211—Selection of the most significant subset of features
- G06F18/2111—Selection of the most significant subset of features by using evolutionary computational techniques, e.g. genetic algorithms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Human Resources & Organizations (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Mathematical Physics (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Biomedical Technology (AREA)
- Genetics & Genomics (AREA)
- Medical Informatics (AREA)
- Acoustics & Sound (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The present invention provides a kind of generation method of pumping plant unit fault diagnosis system, and the data including breaking down to pumping plant unit are collected；Feature vector is extracted to the data information being collected into, and selects optimal characteristics vector from the feature vector of extraction；Optimal characteristics vector and machine learning model are subjected to permutation and combination by genetic algorithm, and select the combination of optimal feature vector and machine learning model；The group of optimal feature vector and machine learning model is combined into diagnostic model to test；Pumping plant unit failure is diagnosed using the diagnostic model being successfully tested, the present invention carries out automation permutation and combination using the genetic algorithm in computer science artificial intelligence field, and select the combination of optimal feature vector and machine learning model, it does not need artificially to be handled and selected, it is at low cost, the period is short, the pumping plant unit diagnostic system formed by the generation method quickly and efficiently detects pumping plant unit failure.
Description
Technical field
The present invention relates to pumping plant technical fields, and in particular to a kind of generation method of pumping plant unit fault diagnosis system.
Background technique
With science and technology and national economy development, pump works agricultural, electric power, petroleum, mining and across flow city water transfer
Etc. in industries using increasingly extensive, especially water drainage, irrigation, power generation, water supply, adjust shipping in terms of to play act foot light
The effect of weight.And water pump assembly is the key equipment of pump works, the superiority and inferiority of operating status directly affects the safe operation of pumping plant,
After China's project of South-to-North water diversion is built up, pumping plant unit long operational time, investment is big, reliability requirement is high, because accident or failure are stopped
Supplying water to interrupt caused by machine will result in significant economic losses and social influence, thus the operation to pumping plant, maintenance, management and
More stringent requirements are proposed for maintenance etc., and especially security reliability has become the important indicator for measuring pumping plant unit quality,
For above situation, the fault diagnosis technology of pumping plant unit comes into being.Fault diagnosis refers in the operation of pumping plant unit, to machine
Group is monitored, and judges whether certain failure occur, to instruct maintenance, help is quickly resumed production.
Existing pumping plant unit fault diagnosis needs to acquire fault data, handles fault data, manually to machine
Learning model and parameter are selected, there are at high cost, the period is long, this to the Function Extension of pumping plant unit fault diagnosis system,
Old algorithm update causes big inconvenience.
Summary of the invention
The present invention provides a kind of generation method of pumping plant unit fault diagnosis system.
In order to solve the above technical problems, the present invention adopts the following technical scheme:
The data to break down to pumping plant unit are collected；
Feature vector is extracted to the data information being collected into, and selects optimal characteristics vector from the feature vector of extraction；
By optimal characteristics vector and machine learning model by genetic algorithm progress permutation and combination, and select optimal feature to
The combination of amount and machine learning model；
The group of optimal feature vector and machine learning model is combined into diagnostic model to test；
Pumping plant unit failure is diagnosed using the diagnostic model for meeting test request.
Further, described to extract feature vector method particularly includes: multiple numbers are converted into collected fault data
Strong point, using data point as the feature vector of the failure.
Further, the selection optimal characteristics vector method particularly includes: calculated according to the feature vector extracted
The variance of itself, while the relevance between feature vector and prediction target being compared, as variance is greater than 1 and feature vector
Relevance between prediction target is high, then this feature vector is optimal characteristics vector, conversely, being then excluded.
Further, described that optimal characteristics vector and machine learning model are subjected to permutation and combination, choosing by genetic algorithm
Select out the combination of optimal feature vector and machine learning model method particularly includes:
S1, N number of initial pool is generated according to optimal feature vector and existing machine learning model at random；
S2, N number of initial pool is generated into N number of new combination by the duplication of genetic algorithm, intersection and variation；
S3, N number of new combination is re-used as the initial pool in S1, and circuits sequentially and executes S2, S3；
S4, optimal N number of combination is obtained after recycling Y times, optimal 1 combination conduct is picked out from optimal N number of combination and is sought
Look for result.
Further, it is described circulation Y time after obtain optimal N number of combination method particularly includes: by be continuously generated newly
Initial pool carries out the iteration of S2, S3, judges whether the number of iterations reaches setting value Y times, is optimal combination after reaching.
Further, the group by optimal feature vector and machine learning model is combined into diagnostic model and tests
Method particularly includes: sample failure to be tested will be needed to diagnose by diagnostic model, the diagnostic result obtained and practical sample
The result of this failure is compared, and then judges whether the accuracy of test is higher than set threshold value, forms diagnosis mould if being higher than
Type, conversely, then backout feature extracts.
Compared with prior art, beneficial effects of the present invention:
The generation method of the pumping plant unit fault diagnosis system using the genetic algorithm in computer science artificial intelligence field into
Row automation permutation and combination, and the combination of optimal feature vector and machine learning model is selected, it does not need manually to machine
Learning model and parameter are selected, at low cost, the period is short, it is good and error is low to calculate effect, so that passing through the generation method
The pumping plant unit diagnostic system of formation can quickly and efficiently detect pumping plant unit failure.
Detailed description of the invention
Fig. 1 is product process structure chart of the present invention；
Fig. 2 is selection optimal diagnosis model flow structure chart.
Specific embodiment
A kind of preferred embodiment of the invention is described in detail with reference to the accompanying drawing.
As shown in Figs. 1-2, a kind of generation method of pumping plant unit fault diagnosis system, comprising:
Step 1: the data to break down to pumping plant unit are collected；
Step 2: being converted into multiple data points to the data information being collected into, mentioned data point as the feature vector of the failure
Take feature vector；
Step 3: calculating the variance of itself to the feature vector of acquisition, while by feature vector and predicting the pass between target
Connection property is compared, and as variance is greater than, the relevance between 1 and feature vector and prediction target is high, then this feature vector is optimal
Feature vector, conversely, being then excluded；
Step 4: automaton learn technology to optimal feature vector and existing machine learning model generate at random it is N number of just
Begin to combine；N number of initial pool is generated into N number of new combination by the duplication of genetic algorithm, intersection and variation；By N number of new group
Conjunction is re-used as initial pool, and generates N number of new combination by the duplication of genetic algorithm, intersection and variation, again again by N
N number of new combination, repetitive cycling above step are generated by genetic algorithm again as initial pool；Whether judge the number of iterations
Reach setting value Y times, is optimal combination after reaching, obtains optimal N number of combination, picked out from optimal N number of combination
1 optimal combination is as searching result；
Step 5: the group of optimal feature vector and machine learning model is combined into diagnostic model, sample event to be tested will be needed
Barrier is diagnosed by diagnostic model, and the diagnostic result obtained is compared with the result of actual sample failure, and then judgement is surveyed
Whether the accuracy of examination is higher than set threshold value, forms diagnostic model if being higher than 80%, conversely, then backout feature extracts, works as return
It is carried out again after feature extraction Step 3: step 4 and step 5, after calculating 3 times repeatedly, the accuracy of such as test is not reach yet
To required threshold value, then return step one, carries out the collection of initial data.
Step 6: being diagnosed using optimal diagnosis model to pumping plant unit failure.
The course of work:
Pump shaft rotor fault: when pump shaft rotor breaks down in pumping plant unit, rotor normal sample quantity, rotor are acquired first
Misalign the throw of sample size, rotor unbalance sample size and the sensor pump shaft Y-direction under contact friction sample size
Signal will form time-domain diagram according to the throw signal of acquisition, then frequency spectrum be obtained using Fourier transform, according to spy from frequency spectrum
Sign rule (this feature rule be extract (0,0.4 ×], (0.4 ×, 0.5 ×), [0.5 ×], (0.5 ×, 1 ×), [1 ×], [2
×], [3 ×, 5 ×], [6 ×, high frequency) this 8 frequency ranges maximum amplitude, wherein " () " represents open interval, " [] " is represented
Closed interval, "×" represent frequency multiplication) extract feature vector, at this time sample data be [Y, 8], (wherein Y representative sample quantity, 8
Represent the characteristic dimension that single sample initial data obtains after feature extraction), each spy is calculated to the feature vector of extraction
The variance of dimension itself is levied, calculation formula is V1=Var(A1, B1 ... ..., X1), in formula: V1 is the variance of dimension 1, and A1 is sample
Dimension 1 that the dimension 1 of notebook data A, B1 are sample data B, the dimension 1 that X1 is sample data X can similarly calculate dimension 2, dimension
Spend 3, dimension 4, dimension 5, dimension 6, the variance of dimension 7 and dimension 8, at the same calculate in feature vector dimension 1 and prediction target it
Between relevance Corr1, comparing formula is Corr1=MI(Vx, P), in formula: MI represents the mutual trust between two variable Vs x and P
Breath, Vx be different sample datas under dimension 1 form feature vector [A1, B1 ... ..., X1], P represent predicted value composition to
It measures [PA, PB ... ..., Px], wherein P1 is predicted value of the sample data A under eight different dimensions, and P2 is sample data B eight
Predicted value under a different dimensions, Px be predicted value of the sample data X under eight different dimensions, can similarly calculate feature to
Relevance in amount between dimension 2, dimension 3, dimension 4, dimension 5, dimension 6, dimension 7, dimension 8 and prediction target, such as each dimension
The relevance that the variance of degree is greater than between 1 and feature vector and prediction target is high, then this feature vector is optimal characteristics vector, instead
It, then be excluded, select accuracy rate in current algebra population preceding 90% machine learning model as machine in next-generation population
Learning model selects two at random from machine learning model, two will selected in obtained optimal characteristics vector and machine
Learning model is intersected and is made a variation by genetic algorithm, is iterated, and an optimal diagnostic model is finally obtained, will be optimal
Diagnostic model be divided into 5 parts, successively select 1 part therein as diagnostic model test sample, be left 4 parts and be used as training sample,
5 training can be carried out to diagnostic model in this way and obtain corresponding 5 diagnostic model test results, with this 5 test results
Average value is carried out as final diagnostic model test result with final diagnostic model test result and true fail result
Compare, form error function, error function is instructed to the training of diagnostic model, sample failure to be tested will be needed to pass through diagnosis mould
Type is diagnosed, and the diagnostic result obtained is compared with the result of actual sample failure, when final model measurement effect (i.e.
Accuracy rate) be higher than minimum frequency 80% when, machine diagnoses pumping plant unit failure using final model measurement, conversely, then
Backout feature extracts, and carries out feature selecting, optimal characteristics extraction and training pattern again after backout feature extracts, calculates repeatedly
After 3 times, if test effect is still below required threshold value, then return step one, carries out the collection of initial data.
Bearing fault: the acceleration sensing when pumping plant unit middle (center) bearing breaks down, first by being mounted on bearings
Device collects the acceleration signal under bearing sample size, wavelet transformation is carried out to acceleration signal, then by first 5 after decomposition
One segment signal is become the vector of one five dimension, to the feature vector of acquisition as the feature vector extracted by wavelet coefficient in this way
Calculate the variance of itself, calculation formula is V1=Var(A1, B1 ... ..., X1), in formula: V1 is the variance of dimension 1, and A1 is sample
Dimension 1 that the dimension 1 of notebook data A, B1 are sample data B, the dimension 1 that X1 is sample data X can similarly calculate dimension 2, dimension
3, dimension 4, the variance of dimension 5 are spent, while calculating the relevance Corr1 in feature vector between dimension 1 and prediction target, are compared
Formula is Corr1=MI(Vx, P), in formula: MI represents the mutual information between two variable Vs x and P, and Vx is under different sample datas
Dimension 1 form feature vector [A1, B1 ... ..., X1], P represent predicted value composition vector [PA, PB ... ..., Px],
Middle P1 is predicted value of the sample data A under five different dimensions, and P2 is prediction of the sample data B under five different dimensions
Value, Px are predicted value of the sample data X under five different dimensions, can similarly calculate dimension 2 in feature vector, dimension 3, dimension
Relevance between degree 4, dimension 5 and prediction target, as variance is greater than the relevance between 1 and feature vector and prediction target
Height, then this feature vector is optimal characteristics vector, conversely, be then excluded, selects in current algebra population accuracy rate preceding 90%
Machine learning model is selected two at random from machine learning model, will be obtained as machine learning model in next-generation population
Optimal characteristics vector and machine in two learning models selecting intersected and made a variation by genetic algorithm, be iterated,
Finally obtain an optimal diagnostic model, optimal diagnostic model be divided into 5 parts, successively select 1 part therein as diagnosis
Model measurement sample is left 4 parts and is used as diagnostic model training samples, can carry out 5 training to diagnostic model in this way and obtain pair
The 5 diagnostic model test results answered, using the average value of this 5 test results as final diagnostic model test result, with
Final diagnostic model test result is compared with true fail result, forms error function, and error function guidance is examined
The training of disconnected model, will need sample failure to be tested to diagnose by diagnostic model, the diagnostic result obtained and practical sample
The result of this failure is compared, and when final model measurement effect (i.e. accuracy rate) is higher than minimum frequency 80%, machine is utilized
Final model measurement diagnoses pumping plant unit failure, conversely, then backout feature extracts, after backout feature extracts again
Feature selecting, optimal characteristics extraction and training pattern are carried out, after calculating 3 times repeatedly, if test effect is still below required threshold value, then
Return step one carries out the collection of initial data.
Cavitation failure: the operation noise near pump intake is collected as fault data, acquires Y sample number of faults
According to since noise is a segment of audio signal, therefore using the mel-frequency cepstrum coefficient of extraction signal as feature vector, the spy
Sign vector one shares 26 dimensions: 12 dimension cepstrum coefficients, 12 dimension cepstrum coefficient difference, 1 dimension energy and 1 dimension energy difference, to acquisition
Feature vector calculate itself variance, calculation formula is V1=Var(A1, B1 ... ..., X1), in formula: V1 is the side of dimension 1
Difference, the dimension 1 that A1 is sample data A, the dimension 1 that B1 is sample data B, the dimension 1 that X1 is sample data X, can similarly calculate
Out dimension 2, dimension 3 ..., the variance of dimension 26, while calculating the relevance in feature vector between dimension 1 and prediction target
Corr1, compare formula for Corr1=MI(Vx, P), in formula: MI represents the mutual information between two variable Vs x and P, and Vx is difference
The feature vector [A1, B1 ... ..., X1] that dimension 1 under sample data forms, P represent predicted value composition vector [PA,
PB ... ..., Px], wherein P1 is predicted value of the sample data A under 26 different dimensions, and P2 is sample data B in 26 differences
Predicted value under dimension, Px are predicted value of the sample data X under 26 different dimensions, can similarly calculate in feature vector and tie up
Degree 2, dimension 3 ..., dimension 26 and prediction target between relevance, as variance be greater than 1 and feature vector and prediction target it
Between relevance it is high, then this feature vector is optimal characteristics vector, conversely, being then excluded, it is accurate in current algebra population to select
Rate is used as machine learning model in next-generation population in preceding 90% machine learning model, selects at random from machine learning model
Two, obtained optimal characteristics vector and two learning models selected in machine are intersected and become by genetic algorithm
It is different, it is iterated, finally obtains an optimal diagnostic model, optimal diagnostic model is divided into 5 parts, is successively selected therein
1 part is used as model measurement sample, is left 4 parts and is used as training sample, can carry out 5 training to model in this way and obtain corresponding 5
A model test results, using the average value of this 5 test results as final diagnostic model test result, with final diagnosis
Model test results are compared with true fail result, form error function, error function is instructed to the instruction of diagnostic model
Practice, sample failure to be tested will be needed to diagnose by diagnostic model, the knot of the diagnostic result obtained and actual sample failure
Fruit is compared, and when final model measurement effect (i.e. accuracy rate) is higher than minimum frequency 80%, machine utilizes final model
Test diagnoses pumping plant unit failure, conversely, then backout feature extracts, carries out feature choosing again after backout feature extracts
Select, optimal characteristics are extracted and training pattern, after calculating 3 times repeatedly, if test effect is still below required threshold value, then return step
One, carry out the collection of initial data.
Embodiment described above is only that preferred embodiments of the present invention will be described, not to model of the invention
It encloses and is defined, without departing from the spirit of the design of the present invention, those of ordinary skill in the art are to technical side of the invention
The various changes and improvements that case is made, should fall within the scope of protection determined by the claims of the present invention.
Claims (6)
1. a kind of generation method of pumping plant unit fault diagnosis model characterized by comprising
The data to break down to pumping plant unit are collected；
Feature vector is extracted to the data information being collected into, and selects optimal characteristics vector from the feature vector of extraction；
By optimal characteristics vector and machine learning model by genetic algorithm progress permutation and combination, and select optimal feature to
The combination of amount and machine learning model；
The group of optimal feature vector and machine learning model is combined into diagnostic model to test；
Pumping plant unit failure is diagnosed using the diagnostic model for meeting test request.
2. the generation method of pumping plant unit fault diagnosis model according to claim 1, which is characterized in that the extraction is special
Levy vector method particularly includes: multiple data points are converted into collected fault data, using data point as the spy of the failure
Levy vector.
3. the generation method of pumping plant unit fault diagnosis model according to claim 1, which is characterized in that the selection is most
Excellent feature vector method particularly includes: calculate the variance of itself according to the feature vector extracted, while by feature vector with
Relevance between prediction target is compared, if variance is greater than the relevance height between 1 and feature vector and prediction target, then
This feature vector is optimal characteristics vector, conversely, being then excluded.
4. the generation method of pumping plant unit fault diagnosis model according to claim 1, which is characterized in that it is described will be optimal
Feature vector and machine learning model select optimal feature vector and machine learning by genetic algorithm progress permutation and combination
The combination of model method particularly includes:
S1, N number of initial pool is generated according to optimal feature vector and existing machine learning model at random；
S2, N number of initial pool is generated into N number of new combination by the duplication of genetic algorithm, intersection and variation；
S3, N number of new combination is re-used as the initial pool in S1, and circuits sequentially and executes S2, S3；
S4, optimal N number of combination is obtained after recycling Y times, optimal 1 combination conduct is picked out from optimal N number of combination and is sought
Look for result.
5. the generation method of pumping plant unit fault diagnosis model according to claim 4, which is characterized in that the circulation Y
Optimal N number of combination is obtained after secondary method particularly includes: the iteration that the new initial pool being continuously generated is carried out to S2, S3 is sentenced
Whether disconnected the number of iterations reaches setting value Y times, is optimal combination after reaching.
6. the generation method of pumping plant unit fault diagnosis model according to claim 1, which is characterized in that it is described will be optimal
Feature vector and the group of machine learning model be combined into what diagnostic model was tested method particularly includes: sample to be tested will be needed
This failure is diagnosed by diagnostic model, and the diagnostic result obtained is compared with the result of actual sample failure, is then sentenced
Whether the accuracy of disconnected test is higher than set threshold value, forms diagnostic model if being higher than, conversely, then backout feature extracts.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910550323.0A CN110399986B (en) | 2019-06-24 | 2019-06-24 | Generating method of pump station unit fault diagnosis system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910550323.0A CN110399986B (en) | 2019-06-24 | 2019-06-24 | Generating method of pump station unit fault diagnosis system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110399986A true CN110399986A (en) | 2019-11-01 |
| CN110399986B CN110399986B (en) | 2023-04-07 |
Family
ID=68323363
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910550323.0A Active CN110399986B (en) | 2019-06-24 | 2019-06-24 | Generating method of pump station unit fault diagnosis system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110399986B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110826737A (en) * | 2019-11-06 | 2020-02-21 | 中水三立数据技术股份有限公司 | Pump station maintenance management method and system based on virtual reality technology |
| CN111126624A (en) * | 2019-12-20 | 2020-05-08 | 北京首汽智行科技有限公司 | Method for judging validity of model prediction result |
| CN112539828A (en) * | 2020-12-08 | 2021-03-23 | 中水三立数据技术股份有限公司 | Pump unit equipment diagnosis method, system and equipment based on curve fitting contrast analysis |
| WO2021203489A1 (en) * | 2020-04-07 | 2021-10-14 | 厦门邑通软件科技有限公司 | Decision behavior recording method, system, and device |
| CN115135358A (en) * | 2020-02-27 | 2022-09-30 | 美国西门子医学诊断股份有限公司 | Automatic sensor tracking verification using machine learning |
| CN117451288A (en) * | 2023-10-16 | 2024-01-26 | 国网经济技术研究院有限公司 | Fault diagnosis method and device for offshore flexible straight platform valve cold main circulating pump |
| CN117556332A (en) * | 2024-01-11 | 2024-02-13 | 北京佰能蓝天科技股份有限公司 | Heating furnace waste heat utilization management method, system and storage medium based on Internet of things |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202952A (en) * | 2016-07-19 | 2016-12-07 | 南京邮电大学 | A kind of Parkinson disease diagnostic method based on machine learning |
| CN108537273A (en) * | 2018-04-08 | 2018-09-14 | 焦点科技股份有限公司 | A method of executing automatic machinery study for unbalanced sample |
| CN108681250A (en) * | 2018-05-14 | 2018-10-19 | 浙江大学 | A kind of improvement machine learning fault diagnosis system based on colony intelligence optimization |
| CN109106384A (en) * | 2018-07-24 | 2019-01-01 | 安庆师范大学 | A kind of psychological pressure condition predicting method and system |
| KR101984248B1 (en) * | 2018-09-13 | 2019-05-30 | 임강민 | Apparatus For Making A Predictive Diagnosis Of Nuclear Power Plant By Machine Learning |
| US20190171950A1 (en) * | 2019-02-10 | 2019-06-06 | Kumar Srivastava | Method and system for auto learning, artificial intelligence (ai) applications development, operationalization and execution |
-
2019
- 2019-06-24 CN CN201910550323.0A patent/CN110399986B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202952A (en) * | 2016-07-19 | 2016-12-07 | 南京邮电大学 | A kind of Parkinson disease diagnostic method based on machine learning |
| CN108537273A (en) * | 2018-04-08 | 2018-09-14 | 焦点科技股份有限公司 | A method of executing automatic machinery study for unbalanced sample |
| CN108681250A (en) * | 2018-05-14 | 2018-10-19 | 浙江大学 | A kind of improvement machine learning fault diagnosis system based on colony intelligence optimization |
| CN109106384A (en) * | 2018-07-24 | 2019-01-01 | 安庆师范大学 | A kind of psychological pressure condition predicting method and system |
| KR101984248B1 (en) * | 2018-09-13 | 2019-05-30 | 임강민 | Apparatus For Making A Predictive Diagnosis Of Nuclear Power Plant By Machine Learning |
| US20190171950A1 (en) * | 2019-02-10 | 2019-06-06 | Kumar Srivastava | Method and system for auto learning, artificial intelligence (ai) applications development, operationalization and execution |
Non-Patent Citations (3)
| Title |
|---|
| PIETER GIJSBERS,ET AL.: "GAMA: Genetic Automated Machine learning Assistant", 《JOURNAL OF OPEN SOURCE SOFTWARE》 * |
| 涂同珩 等: "基于自动机器学习流程优化的雷达辐射源信号识别", 《计算机应用研究》 * |
| 邓森 等: "基于测试性的电子系统综合诊断与故障预测方法综述", 《控制与决策》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110826737A (en) * | 2019-11-06 | 2020-02-21 | 中水三立数据技术股份有限公司 | Pump station maintenance management method and system based on virtual reality technology |
| CN111126624A (en) * | 2019-12-20 | 2020-05-08 | 北京首汽智行科技有限公司 | Method for judging validity of model prediction result |
| CN115135358A (en) * | 2020-02-27 | 2022-09-30 | 美国西门子医学诊断股份有限公司 | Automatic sensor tracking verification using machine learning |
| WO2021203489A1 (en) * | 2020-04-07 | 2021-10-14 | 厦门邑通软件科技有限公司 | Decision behavior recording method, system, and device |
| CN112539828A (en) * | 2020-12-08 | 2021-03-23 | 中水三立数据技术股份有限公司 | Pump unit equipment diagnosis method, system and equipment based on curve fitting contrast analysis |
| CN117451288A (en) * | 2023-10-16 | 2024-01-26 | 国网经济技术研究院有限公司 | Fault diagnosis method and device for offshore flexible straight platform valve cold main circulating pump |
| CN117451288B (en) * | 2023-10-16 | 2024-05-10 | 国网经济技术研究院有限公司 | Fault diagnosis method and device for offshore flexible straight platform valve cold main circulating pump |
| CN117556332A (en) * | 2024-01-11 | 2024-02-13 | 北京佰能蓝天科技股份有限公司 | Heating furnace waste heat utilization management method, system and storage medium based on Internet of things |
| CN117556332B (en) * | 2024-01-11 | 2024-03-15 | 北京佰能蓝天科技股份有限公司 | Heating furnace waste heat utilization management method, system and storage medium based on Internet of things |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110399986B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110399986A (en) | A kind of generation method of pumping plant unit fault diagnosis system | |
| CN108268905A (en) | A kind of Diagnosis Method of Transformer Faults and system based on support vector machines | |
| CN109543210A (en) | A kind of Wind turbines failure prediction system based on machine learning algorithm platform | |
| CN110441065A (en) | Gas turbine online test method and device based on LSTM | |
| CN109034191A (en) | One-dimensional telemetry exception interpretation method based on ELM | |
| CN112799898B (en) | Interconnection system fault node positioning method and system based on distributed fault detection | |
| CN111122162B (en) | Industrial system fault detection method based on Euclidean distance multi-scale fuzzy sample entropy | |
| CN109062189A (en) | A kind of industrial process method for diagnosing faults for complex fault | |
| CN109470946B (en) | Power generation equipment fault detection method and system | |
| CN109187060B (en) | The detection of train speed sensor abnormal signal and axis locking method for diagnosing faults | |
| CN113408068A (en) | Random forest classification machine pump fault diagnosis method and device | |
| CN106647650A (en) | Distributed industrial process monitoring method based variable weighting PCA (Principal Component Analysis) model | |
| CN110060368A (en) | Mechanical method for detecting abnormality based on potential feature coding | |
| CN110647924A (en) | GIS equipment state evaluation method based on support vector description and K-nearest neighbor algorithm | |
| CN112926656A (en) | Method, system and equipment for predicting state of circulating water pump of nuclear power plant | |
| Zhou et al. | Structural health monitoring of offshore wind power structures based on genetic algorithm optimization and uncertain analytic hierarchy process | |
| CN114936520A (en) | Steam turbine fault early warning analysis method based on improved MSET | |
| CN115392782A (en) | Method and system for monitoring and diagnosing health state of process system of nuclear power plant | |
| CN115614292A (en) | Vibration monitoring device and method for vertical water pump unit | |
| CN110378358A (en) | A kind of power distribution network isomeric data integration method and system | |
| Massaferro et al. | Improving electricity non technical losses detection including neighborhood information | |
| Carratù et al. | A novel methodology for unsupervised anomaly detection in industrial electrical systems | |
| CN109543357A (en) | Fault degree quantitative evaluation method for multivariate regression model optimization | |
| CN117216703A (en) | Water delivery pipe network operation data anomaly detection and correction method | |
| CN114046816B (en) | Sensor signal fault diagnosis method based on lightweight gradient lifting decision tree |
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 |