CN110529202A - Steam Turbine Fault Diagnosis detection and method for early warning and system based on power plant's data - Google Patents
Steam Turbine Fault Diagnosis detection and method for early warning and system based on power plant's data Download PDFInfo
- Publication number
- CN110529202A CN110529202A CN201910917039.2A CN201910917039A CN110529202A CN 110529202 A CN110529202 A CN 110529202A CN 201910917039 A CN201910917039 A CN 201910917039A CN 110529202 A CN110529202 A CN 110529202A
- Authority
- CN
- China
- Prior art keywords
- index
- data
- degree
- danger
- steam turbine
- 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
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000003745 diagnosis Methods 0.000 title claims abstract description 24
- 238000001514 detection method Methods 0.000 title claims abstract description 22
- 230000002159 abnormal effect Effects 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000012546 transfer Methods 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims description 31
- 230000005856 abnormality Effects 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 6
- 230000004888 barrier function Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000004088 simulation Methods 0.000 claims description 3
- 230000010355 oscillation Effects 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 abstract description 4
- 238000012544 monitoring process Methods 0.000 description 7
- 239000013598 vector Substances 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 5
- 230000015556 catabolic process Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 4
- 239000003921 oil Substances 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 229910000897 Babbitt (metal) Inorganic materials 0.000 description 1
- 241000580063 Ipomopsis rubra Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000036760 body temperature Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000007418 data mining 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
- 238000000605 extraction Methods 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000010687 lubricating oil Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000010977 unit operation Methods 0.000 description 1
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F01—MACHINES OR ENGINES IN GENERAL; ENGINE PLANTS IN GENERAL; STEAM ENGINES
- F01D—NON-POSITIVE DISPLACEMENT MACHINES OR ENGINES, e.g. STEAM TURBINES
- F01D21/00—Shutting-down of machines or engines, e.g. in emergency; Regulating, controlling, or safety means not otherwise provided for
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F01—MACHINES OR ENGINES IN GENERAL; ENGINE PLANTS IN GENERAL; STEAM ENGINES
- F01D—NON-POSITIVE DISPLACEMENT MACHINES OR ENGINES, e.g. STEAM TURBINES
- F01D21/00—Shutting-down of machines or engines, e.g. in emergency; Regulating, controlling, or safety means not otherwise provided for
- F01D21/003—Arrangements for testing or measuring
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Control Of Turbines (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The Steam Turbine Fault Diagnosis detection and method for early warning and system that the present disclosure proposes a kind of based on power plant's data, carry out that analysis obtains the vibration signal of steam turbine operation achievement data, state shifts index, degree of danger index is input to extreme learning machine training and obtains identification model by obtaining historical data to the historical data of acquisition；Obtain steam turbine operation achievement data in real time simultaneously；The vibration signal of steam turbine operation achievement data, state transfer index, degree of danger index are obtained according to the real-time steam turbine operation achievement data that obtains, above-mentioned overall target is input to trained extreme learning machine to identify, whether output steam turbine operation is abnormal.Few, high reliablity that present disclose provides a kind of variables, the detection of practical shaft system of unit failure and method for early warning, failure exception forecast accuracy is improved, meanwhile, there is exception before the failure occurs in data, it can be detected and be alarmed in advance, avoid detecting bring economic loss afterwards.
Description
Technical field
This disclosure relates to fault diagnosis correlative technology field, in particular to the event of the steam turbine based on power plant's data
Barrier diagnosis detection and method for early warning and system.
Background technique
The statement of this part only there is provided background technical information relevant to the disclosure, not necessarily constitutes first skill
Art.
Electric power plays very important effect in social life and national economy, and steam turbine is that power plant is most important
One of equipment, the monitoring of steamer Axial Status and fault diagnosis are paid high attention to always.Previous steam turbine takes correction maintenance
Mode and plan repair combine, and inefficient and relatively more passive, the initiative of maintenance can be improved in maintenance forecasting.Reinforce vapour
Turbine condition monitoring and fault diagnosis can guarantee the safe operation, prevention and the generation for reducing accident of equipment, and it is hidden to eliminate failure
Suffer from, ensure equipment safety, plays due effect for the development of the national economy.How unit operation in grasp steam turbine operation shape
State finds out possible trouble location and reason, is each correlative study unit and the key content that electric power software company is persistently studied.
In terms of national situations, the Condition Monitoring Technology of the large rotating machineries such as steam turbine has reached quite high level,
A series of condition monitoring system is had developed, and has been successfully applied to production practices.The Steam Turbine vibration prison developed at present
It surveys and fault diagnosis system, the Vibration Condition Monitoring and failure for being mainly used in steamer machine host and subsidiary engine equipment in power plant is examined
It is disconnected, function is realized by obtaining data in real time.These monitoring systems focus on real-time status analysis and diagnosis, functional framework more
Compare similar.
But compared to other monitor system, steam turbine shafting stability special project application both pay close attention to current steam turbine state analysis and
Fault diagnosis, and failure predication can be carried out according to data mining and analysis.There is a large amount of data in existing monitoring system,
These data are mixed and disorderly, not intuitive, feature is unobvious, the data under cover a large amount of knowledge rich ore of " unminding " behind.Existing prison
Examining system " is monitored again and is gently excavated ", payes attention to the abnormality detection based on unalterable rules, and lacks and a large amount of historical datas are worth
Excavation and utilization, exist in terms of the early warning in advance of hidden failures, anticipation obvious insufficient.
Summary of the invention
The disclosure to solve the above-mentioned problems, proposes Steam Turbine Fault Diagnosis detection and the pre- police based on power plant's data
Method and system provide a kind of few variable, high reliablity, the detection of practical shaft system of unit failure and method for early warning, mention
High failure exception forecast accuracy, meanwhile, there is exception before the failure occurs in data, and it can be detected and be alarmed in advance,
It avoids detecting bring economic loss afterwards.
To achieve the goals above, the disclosure adopts the following technical scheme that
One or more embodiments provide the detection of the Steam Turbine Fault Diagnosis based on power plant's data and method for early warning, including
Following steps:
Obtain steam turbine operation achievement data；
It constructs degree of danger function and the danger of each index is obtained according to the achievement data of acquisition and degree of danger function
Degree numerical value；
Steamer is obtained by the degree of danger numerical value of acquisition and according to the degree of danger Weight that history power plant data determine
The degree of danger index at machine system current time；
According to the steam turbine operation achievement data of acquisition judge current time and it is current before all indexs of a certain moment
State transfering state, count each state transfering state index quantity be state shift index；
Vibration signal, state transfer index and the degree of danger index for the steam turbine operation achievement data that will acquire are input to
Trained extreme learning machine is identified whether output steam turbine operation is abnormal.
One or more embodiments provide the detection of the Steam Turbine Fault Diagnosis based on power plant's data and early warning system, packet
It includes:
Data acquisition module: for obtaining steam turbine operation achievement data；
Degree of danger Numerical Simulation Module: for constructing degree of danger function, according to the achievement data of acquisition and dangerous journey
Function is spent, the degree of danger numerical value of each index is obtained；
Degree of danger index determining module: it degree of danger numerical value for that will obtain and is determined according to history power plant data
The degree of danger index at degree of danger Weight acquisition turbine system current time；
State shifts indicator-specific statistics module: judging current time for the steam turbine operation achievement data according to acquisition and works as
The state transfering state of all indexs of a certain moment before preceding, the index quantity for counting each state transfering state is state
Shift index；
Identification module: vibration signal, state transfer index and the danger of the steam turbine operation achievement data for will acquire
Level index is input to trained extreme learning machine and is identified, whether output steam turbine operation is abnormal.
A kind of electronic equipment, the meter run on a memory and on a processor including memory and processor and storage
The instruction of calculation machine when the computer instruction is run by processor, completes step described in the above method.
A kind of computer readable storage medium, for storing computer instruction, the computer instruction is executed by processor
When, complete step described in the above method.
Compared with prior art, the disclosure has the beneficial effect that
(1) disclosure constructs degree of danger index and state turns in the fault detection and prealarming process for carrying out steam turbine
Index is moved, the judge index final as the system failure reduces index quantity, improve the recognition speed of extreme learning machine.Danger
Dangerous level index and state transfer index are to calculate to obtain overall target by the operating index data of all steam turbines, are reduced
The tedious degree of extreme learning machine processing data, improves processing speed, and data cover wide, can reflect current steam turbine
Real-time running state, breakdown judge precision are high.In addition, vibration signal is the important reference of the system failure, directly made
For the direct judge index of the system failure, fault identification precision is further improved.
(2) method of disclosure limit of utilization learning machine explores the relationship between index and system mode, knowledge of testing
Not, the method increase the time efficiency of judgement and accuracys rate.
Detailed description of the invention
The Figure of description for constituting a part of this disclosure is used to provide further understanding of the disclosure, and the disclosure is shown
Meaning property embodiment and its explanation do not constitute the restriction to the disclosure for explaining the disclosure.
Fig. 1 is the method flow diagram according to one or more embodiments；
Fig. 2 is that #4 axis vibration (the 38th variable of label) the failure generation moment of 1 example of the embodiment of the present disclosure occur skyrocketing now
As；
Fig. 3 is that #5 axis vibration (the 39th variable of label) the failure generation moment of 1 example of the embodiment of the present disclosure occur skyrocketing now
As；
Fig. 4 is the network that the achievement data of 1 example of the embodiment of the present disclosure is established；
Fig. 5 is the point intensity of each point in the network of 1 example of the embodiment of the present disclosure.
Specific embodiment:
The disclosure is described further with embodiment with reference to the accompanying drawing.
It is noted that described further below be all exemplary, it is intended to provide further instruction to the disclosure.Unless another
It indicates, all technical and scientific terms used herein has usual with disclosure person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the disclosure.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.It should be noted that not conflicting
In the case where, each embodiment in the disclosure and the feature in embodiment can be combined with each other.Below in conjunction with attached drawing to reality
Example is applied to be described in detail.
Embodiment 1
In the technical solution disclosed in one or more embodiments, as shown in Figure 1, a kind of vapour based on power plant's data
Engine breakdown diagnosis detection and method for early warning, include the following steps:
Step 1 obtains steam turbine operation achievement data in real time；
Step 2, building degree of danger function obtain each index according to the achievement data of acquisition and degree of danger function
Degree of danger numerical value；
Step 3 is obtained by the degree of danger numerical value of acquisition and according to the degree of danger Weight that history power plant data determine
Obtain the degree of danger index at turbine system current time；
Step 4, judged according to the steam turbine operation achievement data of acquisition current time and it is current before a certain moment institute
There is the state transfering state of index, the index quantity for counting each state transfering state shifts index as state；
Step 5, the vibration signal for the steam turbine operation achievement data that will acquire, state shift index, degree of danger index
It is input to trained extreme learning machine to be identified, whether output steam turbine operation is abnormal.
In the fault detection and prealarming process for carrying out steam turbine, in order to reduce index quantity, extreme learning machine is improved
Recognition speed, constructs degree of danger index and state shifts index, the judge index final as the system failure.Degree of danger refers to
Mark and state transfer index are to calculate to obtain overall target by the operating index data of all steam turbines, reduce the limit
Habit machine handles the tedious degree of data, improves processing speed, and data cover wide, can reflect the real time execution of current steam turbine
State, breakdown judge precision are high.In addition, vibration signal is the important reference of the system failure, directly as system event
The direct judge index of barrier, further improves fault identification precision.
Obtaining steam turbine operation achievement data in step 1 in real time may include obtaining turbine inlet steam temperature, steam extraction temperature, taking out
Steam pressure, steam pressure, pressure at expulsion, returning-oil temperature, lubricating oil pressure, oil cooler inlet temperature, oil cooler outlet temperature, axis
Hold vibration, bearing metal temperature, returning-oil temperature, unit load, differential expansion, rotor eccentricity, the indexs such as cylinder is swollen.
It can be sampled with the sampling period of setting, such as can be primary every half a minute acquisition.
Degree of danger function is constructed in step 2, it may be considered that the physical meaning of index constructs the side of degree of danger function
Method, can be with specifically:
21, index is classified: index is divided into two classes, is illustrated respectively in threshold range, the lower numerical value the safer,
For first kind index, the index safer close to average or mode or median refers to for the second class.
22, degree of danger function is established respectively for sorted index: using the lower limit of threshold value as referring to suggestion first
The degree of danger function of class index；Using average as referring to the degree of danger function for suggesting the second class index.
Specifically, first kind index may include vibration class index, the lower Oscillation Amplitude the better；Second class index can wrap
Temperature and pressure index is included, achievement data is not easy too high or too low.
Assuming that first kind index includes n1A index, the second class index include n2A index, degree of danger function are denoted as respectively
Ii(x) and IIj(x).The metrics-thresholds upper and lower bound of first kind index is not to be denoted as l1,iAnd u1,i, average (or mode, or
Median) it is denoted asArtificial threshold interval radius is denoted asSecond class metrics-thresholds upper and lower bound is not remember
Make l2,jAnd u2,j, average (or mode or median) is denoted asArtificial threshold interval radius is denoted as(i
=1 ..., n1, j=1 ..., n2)。
For first kind index, the numerical value of index closer to artificial bottom threshold, it is abnormal a possibility that it is lower, with threshold value
Lower limit is as reference；For the second class index, for the numerical value of index closer to average (or mode or median), abnormal can
Energy property is lower, using average as reference.
The degree of danger function of building is specific as follows:
WhereinWithIndicate index moment t numerical value (i=1 ..., n1, j=1 ..., n2).Degree of danger takes
Value is the degree of danger numerical value I of each indexi(x) and IIj(x) value range is [0,1], and numerical value is bigger, and event occurs for steam turbine
The degree of danger of barrier is higher.
The degree of danger weight of each index and the degree of danger Index of each index can be such that in step 3
(1) degree of danger weight
According to the method for the degree of danger weight that history power plant data determine specifically:
Obtain the data that front and back occurs for failure in history power plant data；The data that front and back occurs for failure are not limited to primary event
Barrier, can be any number of failure.
Step 31, the index for choosing achievement data variation abnormality before and after failure occurs in history power plant data refer to as key
Mark；The data variation can be set as twice or more of variation beyond numerical value when operating normally extremely；It can also be according to event
Hinder data and operate normally the threshold value of the range setting achievement data of data, more than the data that threshold value is variation abnormality, threshold value can
To be rule of thumb arranged, artificial threshold value can also be referred to as.
The abnormal data of key index is abstracted into network by step 32: specifically can by and meanwhile break down pass
Key index connects to form network, and the number that two indices are broken down simultaneously is as the weight on even side；
Step 33, according to calculate network in each point point intensity: for the sum of weight of all adjacent sides of point；
Step 34 calculates corresponding point intensity: for the ratio of the point intensity and frequency of abnormity of changing the time of the point.
The intensities normalised processing of corresponding point is obtained each index degree of danger weight by step 35.
The acquisition process of degree of danger weight can be illustrated by specific example:
Somewhere boiler main fuel trip 21:42:03 on December 1st, 2017 (moment is denoted as tz) tripping, as shown in Fig. 2, #4 axis shakes
(the 38th variable of label) 257um；As shown in figure 3, #5 axis shake (the 39th variable of label) a certain moment before (this when engrave
There is the phenomenon that skyrockets for tq).
The time series chart of 55 indexs is drawn according to historical data, each index of primary part observation is at moment tz and the tq moment
Situation of change.By observation, some index value variation abnormality situations are very serious, some indexs slightly change, some refer to
Mark is almost without variation.There is the case where obvious abnormal and slight abnormality at tz the and tq moment as listed in table 1.
Table 1
In table 1: 1) a, which represents the variable this moment, exception, and numerical value exceeds twice or more of numerical value when operating normally；B generation
Table has exception, but changes less big.2) unlisted or space part indicates the moment without significant change.
1) it executes step 32 and the abnormal data of index is abstracted into network, regard index as node in network, if two
There is exception simultaneously in a index, then connects side between corresponding node.It counts two two indexes while abnormal number occurs, made
Connect the weight on side for two o'clock.A weight matrix can be obtained, and establishes network.Network is as shown in Figure 4.
2) the point intensity of index=all adjacent sides of point the sum of weight reflects following meaning: when occurring abnormal, putting intensity
It is bigger to there is a possibility that abnormal for big index.These indexs are not necessarily most important index, but to system exception ratio
More sensitive index can be referred to as abnormal sensitivity index.If system is likened adult body, these sensitive indicators can be with
It is analogous to body temperature, dry etc., the exception of these indexs will not allow human body crisis occur, but delicately body be reminded to occur
Unsound situation.Index point intensity size shows as shown in Figure 5.
3) there is abnormal number in statistical indicator,Calculated result
As shown in table 2.
Each index point intensity of table 2, accumulative frequency of abnormity and corresponding point intensity
Note: abnormal is with artificial threshold value for reference, more than artificial threshold limits, it is believed that abnormal.Count each moment appearance
The number that abnormal label occurs, as index are accumulative to there is abnormal number.
The corresponding point intensity for calculating acquisition is standardized and is obtained with each index degree of danger weight, specifically
I.e. ai=Dqi/∑Dqi, for the point corresponding point intensity and all the points corresponding point intensity and ratio, wherein aiRefer to for i-th
Target degree of danger weight, DqiFor each index corresponding point intensity.
Some indexs frequently occur the case where superman's work threshold value, but system does not break down；Such as 38 (#4) superthresholds
Number is few, but is once more than, and the probability that system breaks down is very big.Therefore cumulative number is bigger, and weight is smaller.Point intensity
Index and the associated power of other indexs are embodied, associated index is more, and effect is bigger, therefore point intensity is bigger, weight
It is bigger.Therefore using the standardization of corresponding point intensity as weight.
(2) degree of danger index is calculated
It can all affect to system since exception occurs in each index, but the degree of each Index Influence is not
Together, can obtain the overall target of index using method of weighting is degree of danger index, and degree of danger index can indicate are as follows:
Wherein, A1It is the vector that first kind degree of danger weight is sequentially arranged composition, A2It is the second class index degree of danger power
It is sequentially arranged the vector of composition again,It is calculated and is obtained according to the numerical value of the first kind index of the acquisition of moment t
Degree of danger numerical value,The danger obtained is calculated according to the numerical value of the second class index of the acquisition of moment t
Degree numerical value.
Step 4 is used to determine another index of breakdown judge: state shifts index, and the determination method that state shifts index can
With specific as follows
With the threshold range that sets as reference, state transfering state include achievement data from threshold range variation be super
Out threshold value, achievement data always in threshold range, achievement data is from being in threshold range and index number beyond changes of threshold
According to exceeding four kinds of states of threshold value always.
Using artificial threshold value as reference, numerical value is fallen within a threshold range, and is marked " 1 ", is otherwise labeled as " 0 ".For some
For index, state transfering state from moment t to moment t+1 four kinds, respectively 0 → 0 nothing but, 1 → 0,0 → 1,1 → 1.Such as
Fruit index persistently occurs 0 → 0,1 → 0, then the Indexes Abnormality situation is more obvious than 0 → 1,1 → 1.
Statistics various states from moment t to t+1 shift quantity.State transfering state can indicate are as follows: ss (t) is such as enabled to indicate
From 1 → 1 index quantity, si (t) indicates the index quantity from 1 → 0, and is (t) indicates the index quantity from 0 → 1, ii (t) table
Show the index quantity from 0 → 0.
The vibration signal for the steam turbine operation achievement data that will acquire in step 5, state shift index, degree of danger index
It is input to trained extreme learning machine to be identified, whether output steam turbine operation is abnormal.
The vibration signal of steam turbine operation achievement data includes the vibration signal of bearing, can directly extract vibration signal
Amplitude data, the input as extreme learning machine.
It being identified in the present embodiment using extreme learning machine, extreme learning machine is a kind of novel neural network algorithm,
Can preferably approach complicated nonlinear function, the algorithm to the input weight of network and offset into row stochastic assignment,
Without passing through gradient descent algorithm iteration adjustment as Single hidden layer feedforward neural networks, therefore reduce many Human disturbances.
Algorithm is as follows:
There is N number of different sample (xi,ti), wherein xi=[xi1,xi2,...,xin]T∈Rn,ti=[ti1,ti2,...,tim]
∈Rm, L hidden node is contained in network, activation primitive is G (x), then formula can be expressed as follows:
Wherein j=1 ..., N, ai=(ai1,ai2,...,ain)TFor input layer node xjWith the connection of i-th of hidden layer node
Weight vector；βi=(βi1,βi2,...,βin) indicate i-th of hidden layer Node connectedness output layer neuron weight vector；bjIt is hidden
The offset of layer.aixjIndicate aiWith xjInner product, aiAnd biRespectively indicate i-th of radial basis function node center and influence because
Son, or the connection weight and deviation that are expressed as between input layer and hidden layer.
The neural network contains three layers: input layer, hidden layer and output layer.Wherein hidden layer includes L hidden neuron, if hidden
Layer nodal point number L=N, then above formula can be write as matrix form: H β=T, wherein network hidden layer output matrix are as follows:
If neural network can error-free prediction training sample, the weight of hidden layer and output layer is that have solution
, L=N at this time, H is reversible the square matrix of N × N, but actually L is often much smaller than N, then the problem of solving weight vector is
There is error, i.e., there are errors between network output and actual value.The training of neural network is just to solve for linear system H β in fact
=T, extreme learning machine target be minimize training error | | H β-T | | and and acquire hidden layer output weight vectors β, standardization
Least square method is for solving linear system:Here H-1It is the generalized inverse matrix of H, therefore predicted value Y can
To indicate are as follows:So error e can be expressed as following formula: e=| | Y-T | |2=| | HH-1T-T||2
The training method of the extreme learning machine of the present embodiment are as follows:
Step 51, the history power plant data for obtaining steam turbine operation achievement data, including fault data and normal data；
Step 52 calculates the degree of danger weight for determining index according to history power plant data；It is determined in this step and step 3
The method of the degree of danger weight of each index is identical.
Step 53, building degree of danger function obtain each index according to the achievement data of acquisition and degree of danger function
Degree of danger numerical value；This step is identical as the calculation method of the building of degree of danger function and degree of danger numerical value in step 3.
Step 54 is obtained by the degree of danger numerical value of acquisition and according to the degree of danger Weight that history power plant data determine
Obtain the degree of danger index at turbine system current time；This step is identical as the calculation method of degree of danger index in step 3.
Step 55 calculates the state that all indexs of front and back or any two time data operated normally occur for failure
Transfering state counts the index quantity of each state transfering state；The determination side of state transfer index in this step and step 4
Method is identical.
Step 56 is made with the vibration signal of operating index data, the index quantity of state transfering state, degree of danger index
It is output with the operating status of steam turbine for input, to extreme learning machine training and test, obtains the parameter of extreme learning machine.
Using above method training extreme learning machine, randomly selects in tables of data 500,000 and be used as training sample, residue nearly 6
Ten thousand datas are as test sample.Selecting the training number of plies is 3 layers, and hidden neuron number 300, training function is ' sig '.Instruction
Practice time 3.447s, trained and test accuracy rate reaches 99.94% or more.If the sampling interval is 5 seconds, sampled data is within one day
24*60*12=17280 item.False Rate is 17280* (1-99.94%)=10.36 < 11.By analysis, erroneous judgement situation is
Normal data is judged as that wrong data, wrong data can judge.The training time of the algorithm is 3.4 seconds, and training is quasi-
True rate is 99.94%, test accuracy rate 99.94%, and real-time with higher and generalization ability are applied to actual power plant
The Steam Turbine Fault Diagnosis of data, before steam turbine breaks down, each achievement data is generally just abnormal variation, is by this
The method of system may be implemented to give warning in advance, and improves the stability of the operation of steam turbine and improves the service life of steam turbine.
Embodiment 2
The present embodiment provides a kind of Steam Turbine Fault Diagnosis detection and early warning system based on power plant's data, comprising:
Data acquisition module: for obtaining steam turbine operation achievement data；
Degree of danger Numerical Simulation Module: for constructing degree of danger function, according to the achievement data of acquisition and dangerous journey
Function is spent, the degree of danger numerical value of each index is obtained；
Degree of danger index determining module: it degree of danger numerical value for that will obtain and is determined according to history power plant data
The degree of danger index at degree of danger Weight acquisition turbine system current time；
State shifts indicator-specific statistics module: judging current time for the steam turbine operation achievement data according to acquisition and works as
The state transfering state of all indexs of a certain moment before preceding, the index quantity for counting each state transfering state is state
Shift index；
Identification module: the vibration signal of the steam turbine operation achievement data for will acquire, state transfer index, dangerous journey
Degree index is input to trained extreme learning machine and is identified, whether output steam turbine operation is abnormal.
Further include: extreme learning machine training module: for training extreme learning machine.
Embodiment 3
The present embodiment provides a kind of electronic equipment, including memory and processor and storage on a memory and are being handled
The computer instruction run on device when the computer instruction is run by processor, completes step described in the method for embodiment 1
Suddenly.
Embodiment 4
The present embodiment provides a kind of computer readable storage mediums, for storing computer instruction, the computer instruction
When being executed by processor, step described in 1 method of embodiment is completed.
The foregoing is merely preferred embodiment of the present disclosure, are not limited to the disclosure, for the skill of this field
For art personnel, the disclosure can have various modifications and variations.It is all within the spirit and principle of the disclosure, it is made any to repair
Change, equivalent replacement, improvement etc., should be included within the protection scope of the disclosure.
Although above-mentioned be described in conjunction with specific embodiment of the attached drawing to the disclosure, model not is protected to the disclosure
The limitation enclosed, those skilled in the art should understand that, on the basis of the technical solution of the disclosure, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within the protection scope of the disclosure.
Claims (10)
1. Steam Turbine Fault Diagnosis detection and method for early warning based on power plant's data, characterized in that include the following steps:
Obtain steam turbine operation achievement data；
It constructs degree of danger function and the degree of danger of each index is obtained according to the achievement data of acquisition and degree of danger function
Numerical value；
Steam turbine system is obtained by the degree of danger numerical value of acquisition and according to the degree of danger Weight that history power plant data determine
The degree of danger index at system current time；
According to the steam turbine operation achievement data of acquisition judge current time and it is current before all indexs of a certain moment shape
State transfering state, the index quantity for counting each state transfering state is that state shifts index；
Vibration signal, state transfer index and the degree of danger index for the steam turbine operation achievement data that will acquire are input to training
Good extreme learning machine is identified whether output steam turbine operation is abnormal.
2. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as described in claim 1, it is characterized in that: pole
The training method for limiting learning machine, includes the following steps:
Obtain the history power plant data of steam turbine operation achievement data, including fault data and normal data；
The degree of danger weight for determining index is calculated according to history power plant data；
It constructs degree of danger function and the degree of danger of each index is obtained according to the achievement data of acquisition and degree of danger function
Numerical value；
Steam turbine system is obtained by the degree of danger numerical value of acquisition and according to the degree of danger Weight that history power plant data determine
The degree of danger index at system current time；
Calculate the state transfering state that all indexs of front and back or any two time data operated normally occur for failure, system
The index quantity for counting each state transfering state is that state shifts index；
Using the vibration signal of operating index data, state transfer index and degree of danger index as input, with the fortune of steam turbine
Row state is output, to extreme learning machine training and test, obtains the parameter of extreme learning machine.
3. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as claimed in claim 1 or 2, feature
It is: according to the method for the degree of danger weight that history power plant data determine, comprising:
Step 31 chooses the index of achievement data variation abnormality before and after failure occurs in history power plant data as key index；
The abnormal data of key index is abstracted into network by step 32: specifically by and meanwhile break down key index connect
It connects to form network, the number that two indices are broken down simultaneously is as the weight on even side；
Step 33, according to calculate network in each point point intensity: for the sum of weight of all adjacent sides of point；
Step 34 calculates corresponding point intensity: for the ratio of the point intensity and frequency of abnormity of changing the time of the point.
The intensities normalised processing of corresponding point is obtained each index degree of danger weight by step 35.
4. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as claimed in claim 3, characterized in that therefore
The determination method of front and back achievement data variation abnormality occurs for barrier are as follows: front and back achievement data occurs for failure beyond numerical value when operating normally
Twice or twice or more；
Or
Given threshold, failure occur front and back achievement data variation and exceed threshold range.
5. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as claimed in claim 1 or 2, feature
It is:
The method for constructing degree of danger function, specifically:
Index is classified: in threshold range, more lower safer index is first kind index to numerical value；In threshold range
Interior, the index safer close to average or mode or median is the second class index；
Degree of danger function is established respectively for sorted index: using the lower limit of threshold value as referring to suggestion first kind index
Degree of danger function；Using average as referring to the degree of danger function for suggesting the second class index.
6. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as claimed in claim 5, it is characterized in that: the
A kind of index includes vibration class index, and the lower Oscillation Amplitude the better；Second class index includes temperature and pressure index, achievement data
It is not easy too high or too low.
7. the Steam Turbine Fault Diagnosis detection based on power plant's data and method for early warning as described in claim 1, it is characterized in that: shape
State transfering state include achievement data from threshold range variation for beyond threshold value, achievement data always in threshold range,
From exceeding, changes of threshold is in threshold range to achievement data and achievement data is exceeding always four kinds of states of threshold value.
8. Steam Turbine Fault Diagnosis detection and early warning system based on power plant's data, characterized in that include:
Data acquisition module: for obtaining steam turbine operation achievement data；
Degree of danger Numerical Simulation Module: for constructing degree of danger function, according to the achievement data of acquisition and degree of danger letter
Number, obtains the degree of danger numerical value of each index；
Degree of danger index determining module: degree of danger numerical value and the danger determining according to history power plant data for that will obtain
The degree of danger index at degree Weight acquisition turbine system current time；
State shifts indicator-specific statistics module: judging current time and currently it for the steam turbine operation achievement data according to acquisition
The state transfering state of preceding all indexs of a certain moment counts the index quantity of each state transfering state as state transfer
Index；
Identification module: vibration signal, state transfer index and the degree of danger of the steam turbine operation achievement data for will acquire
Index is input to trained extreme learning machine and is identified, whether output steam turbine operation is abnormal.
9. a kind of electronic equipment, characterized in that on a memory and on a processor including memory and processor and storage
The computer instruction of operation when the computer instruction is run by processor, is completed described in any one of claim 1-7 method
Step.
10. a kind of computer readable storage medium, characterized in that for storing computer instruction, the computer instruction is located
When managing device execution, step described in any one of claim 1-7 method is completed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910917039.2A CN110529202B (en) | 2019-09-26 | 2019-09-26 | Steam turbine fault diagnosis detection and early warning method and system based on power plant data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910917039.2A CN110529202B (en) | 2019-09-26 | 2019-09-26 | Steam turbine fault diagnosis detection and early warning method and system based on power plant data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110529202A true CN110529202A (en) | 2019-12-03 |
| CN110529202B CN110529202B (en) | 2020-07-28 |
Family
ID=68670385
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910917039.2A Active CN110529202B (en) | 2019-09-26 | 2019-09-26 | Steam turbine fault diagnosis detection and early warning method and system based on power plant data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110529202B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111339074A (en) * | 2020-02-24 | 2020-06-26 | 深圳市名通科技股份有限公司 | Threshold generation method, device, equipment and storage medium |
| CN111383428A (en) * | 2020-05-29 | 2020-07-07 | 成都千嘉科技有限公司 | Online meter state monitoring and early warning method and system |
| CN111445105A (en) * | 2020-02-28 | 2020-07-24 | 山东电力工程咨询院有限公司 | Power plant online performance diagnosis method and system based on target value analysis |
| CN112017793A (en) * | 2020-08-28 | 2020-12-01 | 中国科学院合肥物质科学研究院 | Molecular pump maintenance decision management system and method for fusion device |
| CN112541563A (en) * | 2020-09-30 | 2021-03-23 | 国电龙源电力技术工程有限责任公司 | Rotary equipment vibration prediction management system based on edge calculation technology |
| CN114396323A (en) * | 2022-01-18 | 2022-04-26 | 中电华创电力技术研究有限公司 | Intelligent early warning method and device for cylinder uncovering maintenance of steam turbine body |
| CN115062653A (en) * | 2022-06-09 | 2022-09-16 | 山东龙源电力工程有限公司 | Analysis maintenance system based on steam turbine of thermal power plant |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107036819A (en) * | 2017-05-02 | 2017-08-11 | 大唐东北电力试验研究所有限公司 | The Turbo-generator Set remote oscillation fault diagnosis method and system of multi-parameter amendment |
| CN109033719A (en) * | 2018-09-12 | 2018-12-18 | 温州大学苍南研究院 | A kind of wind turbine Method for Bearing Fault Diagnosis |
| CN109299582A (en) * | 2018-12-03 | 2019-02-01 | 黑龙江苑博信息技术有限公司 | Steam turbine sliding pressure optimization of profile method based on unit operation big data multidimensional ordering |
-
2019
- 2019-09-26 CN CN201910917039.2A patent/CN110529202B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107036819A (en) * | 2017-05-02 | 2017-08-11 | 大唐东北电力试验研究所有限公司 | The Turbo-generator Set remote oscillation fault diagnosis method and system of multi-parameter amendment |
| CN109033719A (en) * | 2018-09-12 | 2018-12-18 | 温州大学苍南研究院 | A kind of wind turbine Method for Bearing Fault Diagnosis |
| CN109299582A (en) * | 2018-12-03 | 2019-02-01 | 黑龙江苑博信息技术有限公司 | Steam turbine sliding pressure optimization of profile method based on unit operation big data multidimensional ordering |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111339074A (en) * | 2020-02-24 | 2020-06-26 | 深圳市名通科技股份有限公司 | Threshold generation method, device, equipment and storage medium |
| CN111339074B (en) * | 2020-02-24 | 2023-05-05 | 深圳市名通科技股份有限公司 | Threshold generation method, device, equipment and storage medium |
| CN111445105A (en) * | 2020-02-28 | 2020-07-24 | 山东电力工程咨询院有限公司 | Power plant online performance diagnosis method and system based on target value analysis |
| CN111445105B (en) * | 2020-02-28 | 2023-07-14 | 山东电力工程咨询院有限公司 | Power plant online performance diagnosis method and system based on target value analysis |
| CN111383428A (en) * | 2020-05-29 | 2020-07-07 | 成都千嘉科技有限公司 | Online meter state monitoring and early warning method and system |
| CN112017793A (en) * | 2020-08-28 | 2020-12-01 | 中国科学院合肥物质科学研究院 | Molecular pump maintenance decision management system and method for fusion device |
| CN112017793B (en) * | 2020-08-28 | 2021-09-03 | 中国科学院合肥物质科学研究院 | Molecular pump maintenance decision management system and method for fusion device |
| CN112541563A (en) * | 2020-09-30 | 2021-03-23 | 国电龙源电力技术工程有限责任公司 | Rotary equipment vibration prediction management system based on edge calculation technology |
| CN114396323A (en) * | 2022-01-18 | 2022-04-26 | 中电华创电力技术研究有限公司 | Intelligent early warning method and device for cylinder uncovering maintenance of steam turbine body |
| CN114396323B (en) * | 2022-01-18 | 2024-04-26 | 中电华创电力技术研究有限公司 | Intelligent early warning method and device for overhauling uncovering cylinder of turbine body |
| CN115062653A (en) * | 2022-06-09 | 2022-09-16 | 山东龙源电力工程有限公司 | Analysis maintenance system based on steam turbine of thermal power plant |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110529202B (en) | 2020-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110529202A (en) | Steam Turbine Fault Diagnosis detection and method for early warning and system based on power plant's data | |
| Hsu et al. | Wind turbine fault diagnosis and predictive maintenance through statistical process control and machine learning | |
| Zhi-Ling et al. | Expert system of fault diagnosis for gear box in wind turbine | |
| Wang et al. | An overview of industrial alarm systems: Main causes for alarm overloading, research status, and open problems | |
| CN102721901B (en) | Based on the electric network failure diagnosis method of sequential Bayes knowledge base TBKB | |
| CN105376193B (en) | The intelligent association analysis method and device of security incident | |
| CN110322018A (en) | A kind of power plant fans fault early warning system based on fuzzy reasoning | |
| CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
| CN107942994A (en) | A kind of satellite temperature control system method for diagnosing faults based on temperature curve feature | |
| CN106872172A (en) | The method for real time discriminating and system of Aero Engine Testing security parameter monitoring | |
| CN110337640B (en) | Methods, systems, and media for problem alert aggregation and identification of suboptimal behavior | |
| Holbert et al. | Nuclear power plant instrumentation fault detection using fuzzy logic | |
| CN111563685B (en) | Power generation equipment state early warning method based on auto-associative kernel regression algorithm | |
| CN109325553A (en) | A kind of wind turbine gearbox fault detection method, system, equipment and medium | |
| Nayeri et al. | Fault detection and isolation of gas turbine: Hierarchical classification and confidence rate computation | |
| CN108074045A (en) | The vulnerability analysis of Wind turbines complete machine and fault sequencing method and electric terminal | |
| Ohana et al. | Data-mining experiments on a hydroelectric power plant | |
| Arpishkin et al. | Intelligent integrity monitoring system for technological process data | |
| CN107290665A (en) | A kind of diagnostic system and method for thermal power generation unit Water vapor chemistry failure | |
| Liu et al. | DRES: Data recovery for condition monitoring to enhance system reliability | |
| Stamatis | Engine condition monitoring and diagnostics | |
| Yang et al. | Evaluation method of rotating machinery health state based on TPE-XGBoost | |
| Gelman et al. | Novel anomaly detection technique based on the nearest neighbour and sequential methods | |
| CN114893428B (en) | Fine pre-diagnosis and accurate operation and maintenance device and system for ventilator | |
| Li et al. | A deep neural network based fault diagnosis method for centrifugal chillers |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240429 Address after: No. 2226, 22nd Floor, No. 66 Cuilong Street, Longquan Street, Longquanyi District, Chengdu City, Sichuan Province, 610000 Patentee after: Sichuan Panyingda Technology Co.,Ltd. Country or region after: China Address before: 250353 University Road, Changqing District, Ji'nan, Shandong Province, No. 3501 Patentee before: Qilu University of Technology Country or region before: China |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240430 Address after: 152000 Dongtaiping Eighth Committee, Zhaodong City, Suihua City, Heilongjiang Province Patentee after: Zhaodong Huaqing New Energy Co.,Ltd. Country or region after: China Address before: No. 2226, 22nd Floor, No. 66 Cuilong Street, Longquan Street, Longquanyi District, Chengdu City, Sichuan Province, 610000 Patentee before: Sichuan Panyingda Technology Co.,Ltd. Country or region before: China |
|
| TR01 | Transfer of patent right |