CN102999038B - The diagnostic device of generating set and the diagnostic method of generating set - Google Patents
The diagnostic device of generating set and the diagnostic method of generating set Download PDFInfo
- Publication number
- CN102999038B CN102999038B CN201210288454.4A CN201210288454A CN102999038B CN 102999038 B CN102999038 B CN 102999038B CN 201210288454 A CN201210288454 A CN 201210288454A CN 102999038 B CN102999038 B CN 102999038B
- Authority
- CN
- China
- Prior art keywords
- measuring
- generating set
- service condition
- signal
- diagnostic
- 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.)
- Active
Links
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
A kind of diagnostic device, cannot detect abnormal failing to report when being reduced in abnormal generation, diagnostic accuracy is high.The running status of diagnostic device is carried out based on the measuring-signal measured from generating set measuring state, diagnostic device diagnostic result being shown in the generating set of image display device possesses: data mapping unit, is used in the measuring-signal that the quantity of state of measuring generating set in the diagnostic device of generating set obtains and constructs the model used in diagnosis; Model definition unit, defines the method for normalizing carrying out service condition and the measuring-signal diagnosed with described model; Diagnosis unit, diagnoses the running status of generating set with the model constructed by described data mapping unit; Possess in described model definition unit and judge the service condition detection unit of service condition of generating set and the normalizing condition determination section of the normalizing condition by each service condition determination data judged by service condition detection unit, in described diagnosis unit, mate service condition diagnose to switch diagnostic model.
Description
Technical field
The present invention relates to the diagnostic device of generating set (plant) and the diagnostic method of generating set.
Background technology
When the diagnostic device of generating set produces abnormal transient or accident in a device, detect the generation of this exception or accident according to the measuring-signal from equipment.
As the diagnostic device of the equipment of known case, in Japanese Unexamined Patent Publication 2005-165375 publication, disclose the diagnostic device of the equipment using self-elevating platform ART (AdaptiveResonanceTheory:ART).At this, ART is the technology data of multidimensional being categorized as classification according to its similarity.
In the technology of the diagnostic device of the equipment described in this JP 2005-165375 publication, first, the state carrying out extraction equipment according to the measuring-signal in past of the service data that have recorded equipment is considered to the measuring-signal of normal period, is used as data mapping data.Then, use ART, be multiple classification (normal category) by data mapping Data classification, make the normal model used in diagnosis.Next, with ART, the current measuring-signal of equipment is categorized as classification.Measuring-signal in this prior and normal model inconsistent time, when namely cannot be categorized as normal category, generate new classification (new classification).That is, the generation of new classification means that the tendency of measuring-signal changes, and the state of equipment there occurs change.Therefore, this judges abnormal generation with the generation of new classification, is diagnosed as abnormal technology when the generation ratio of new classification has exceeded threshold value.
Patent documentation
Patent documentation 1:JP JP 2005-165375 publication
Generating set starting, stopping, constant load, load change etc. run all under various conditions.If the condition run is different, then the scope of the change of measuring-signal is also different.
In addition, as the pre-treatment of the measuring-signal used in diagnosis, it is normalized.In normalized, normalized lower limit is set to 0, normalized higher limit is set to 1, so process measuring-signal.Normalized lower limit and higher limit need to set in advance.In the diagnostic device of prior art, no matter service condition how, all under identical normalizing condition, processes measuring-signal.Therefore, need normalization scope to be set to wider scope, to make the variation range of the measuring-signal that can comprise under whole service conditions.
If normalization scope is larger than the variation range of measuring-signal, then the change of the value after being normalized by measuring-signal will diminish.Therefore, the tendency change of measuring-signal when extremely occurring cannot be caught, also cannot produce new classification even if exist when abnormal generation.This becomes the reason failed to report.
Summary of the invention
The object of the present invention is to provide one suitably to determine normalization scope by mating with service condition, suppressing the diagnostic device failing to report, improve diagnostic accuracy thus.
The diagnostic device of generating set, the running status of diagnostic device is carried out based on the measuring-signal obtained from generating set measuring state amount, and diagnostic result is shown in image display device, it is characterized in that, possess: data mapping unit, it is used in the quantity of state measuring generating set in the diagnostic device of generating set and the measuring-signal obtained, and constructs the model used in diagnosis; Model definition unit, the method for normalizing of service condition and the measuring-signal diagnosed is carried out in its definition by described model; And diagnosis unit, it uses the model constructed by described data mapping unit to diagnose the running status of generating set; Possess in described model definition unit: service condition detection unit, it judges the service condition of generating set; With normalizing condition determination section, it decides the normalizing condition of data by each service condition judged by service condition detection unit, in described diagnosis unit, mates switch diagnostic model and diagnose with service condition.
The diagnostic device of the generating set of the application of the invention, cannot detect abnormal failing to report when can reduce abnormal generation, improve diagnostic accuracy.In addition, can automatically determine normalization scope, can shorten diagnostic device adjustment period between.
Accompanying drawing explanation
Fig. 1 is the control block diagram of the formation of the diagnostic device of the generating set representing one embodiment of the present of invention.
Fig. 2 is the process flow diagram of the elemental motion of the diagnostic device representing the generating set shown in Fig. 1 and represents action key diagram regularly.
Fig. 3 is the key diagram to the installation example of the function that data are classified in the data mapping unit represented in the diagnostic device of the equipment shown in Fig. 1, diagnosis unit.
Fig. 4 is the key diagram of the example that measuring-signal is classified by the data mapping unit in the diagnostic device of the generating set represented as shown in Figure 1.
Fig. 5 is the key diagram of the relation representing originate mode in the generating set shown in Fig. 4 (a) and process values.
Fig. 6 is the key diagram of the action of service condition detection unit 500 in the diagnostic device of the generating set shown in Fig. 1.
Fig. 7 is the key diagram of the 1st embodiment of normalizing condition determination section 600 in the diagnostic device of the generating set shown in Fig. 1.
Fig. 8 is the key diagram of the 2nd embodiment of normalizing condition determination section 600 in the diagnostic device of the generating set shown in Fig. 1.
Fig. 9 is the key diagram of the 3rd embodiment of normalizing condition determination section 600 in the diagnostic device of the generating set shown in Fig. 1.
Figure 10 is the key diagram of the motion flow of data mapping pattern in the diagnostic device of the generating set shown in Fig. 1 and diagnostic mode.
Figure 11 is the key diagram of the data shape be kept in the database in the diagnostic device of the generating set shown in Fig. 1.
Figure 12 is the key diagram of the effect of the diagnostic device of the generating set shown in Fig. 1.
Figure 13 is the key diagram of the picture be shown on the image display device in the diagnostic device of the generating set shown in Fig. 1.
Symbol description:
1,3,4 measuring-signals
2 external input signals
5,6,7 model definition information
8,9 model informations
10 diagnostic results
11 picture display information
50 diagnostic device information
100 equipment
200 diagnostic devices
210 outer input interfaces
220 outside output interfaces
310 measuring-signal databases
320 pattern definition number are according to storehouse
330 diagnostic model databases
400 model definition units
500 service condition detection units
600 normalizing condition determination sections
700 data mapping unit
800 diagnosis units
900 operational management rooms
910 external input device
920 keyboards
930 mouses
940 image display devices
Embodiment
Next, the diagnostic device of the generating set as embodiments of the invention is described below with reference to accompanying drawing.
Fig. 1 is the block diagram of the diagnostic device of the generating set illustrated as one embodiment of the present of invention.In the diagnostic device of the generating set shown in Fig. 1, carried out the state of diagnostic device 100 by diagnostic device 200.
Diagnostic device 200 possesses model definition unit 400, data mapping unit 700 and diagnosis unit 800, is used as the arithmetic unit forming diagnostic device 200.This diagnostic device 200 possesses measuring-signal database 310, pattern definition number according to storehouse 320 and diagnostic model database 330, is used as database.In addition, in FIG, database is slightly designated as DB.
In measuring-signal database 310, pattern definition number are according to the database of storehouse 320 and diagnostic model database 330, record the information of electronization, usually they are called e-file (electronic data).
Data mapping unit 700, according to the measuring-signal of the running status of measuring equipment 100, based on the savings data of measuring-signal of running status in past having put aside equipment 100, is made the diagnostic model of the normal condition of the equipment of have learned.
The method for normalizing of model definition unit 400 to the measuring-signal when service condition diagnosed with diagnostic model and data mapping defines.
The data of the value of the diagnostic model be made by data mapping unit 700 and the data of the measuring-signal of equipment 100 measured compare by diagnosis unit 800.If measuring-signal is consistent with the value of the diagnostic model that have learned normal condition, then the state of equipment is judged to normally, if inconsistent, to be judged to be exception.
In addition, diagnostic device 200 possesses outer input interface 210 and outside output interface 220, is used as the interface with outside.
Then, via outer input interface 210, namely the running status of equipment 100 is measured various quantity of state and the measuring-signal 1 obtained and by the operation of the external input device 910 be made up of keyboard 920 and mouse 930 that possesses in operational management room 900 and the external input signal 2 be made be taken in diagnostic device 200.In addition, image display information 11 exported to via outside output interface 220 image display device 940 that operational management room 900 possesses.
In addition, in the diagnostic device of the equipment of the present embodiment, model definition unit 400, data mapping unit 700, diagnosis unit 800, measuring-signal database 310, pattern definition number is possessed according to storehouse 320, diagnostic model database 330 in the inside of diagnostic device 200, but also their part can be configured at the outside of diagnostic device 200, only data be communicated between the devices.
In addition, show in the diagnostic device of the equipment of the present embodiment, the equipment 100 as diagnosis object is the situation of 1, but also can diagnose many seat apparatus 100 with 1 diagnostic device 200.
Next, the action of the diagnostic device 200 possessed in the diagnostic device of the equipment of the present embodiment is described.
In the diagnostic device of the equipment of the present embodiment shown in Fig. 1, via outer input interface 210, the measuring-signal 1 be measured the various quantity of states of equipment 100 is taken into.Measuring-signal 3 is stored in the measuring-signal database 310 be arranged in diagnostic device 200.
Service condition detection unit 500 and normalizing condition determination section 600 is possessed respectively in model definition unit 400.Model definition information 5, relative to the input of measuring-signal 4, is exported to pattern definition number according to storehouse 320 by model definition unit 400.
Generating set has makes that the constant constant load run of output is run, the load change that is out of service, that make exporting change of the starting operation of starting equipment, arrestment runs.In service condition detection unit 500, to be used in measuring-signal database 310 the savings data of the equipment 100 of savings, by these Data Segmentations to be during constant load is run, in starting operation, out of service in, load changes operating operational mode.In addition, the characteristic quantity of respective operational mode is extracted.Use Fig. 6 that the detailed of this function is described.
In addition, in data mapping unit 700, measuring-signal is normalized, as the pre-treatment for constructing model.In normalizing condition determination section 600, decide suitable normalizing condition for each operational mode.Fig. 7 ~ Fig. 9 is used to describe the detailed of this function later.Aforesaid model definition information 5 is made up of service condition data and normalizing condition data.
In data mapping unit 700, be used in the measuring-signal 4 of the equipment 100 put aside in measuring-signal database 310 and be kept at pattern definition number according to the model definition information 6 in storehouse 320, constructing the model used in diagnosis.The model information 8 be made by data mapping unit 700 is kept in diagnostic model database 330.
As the technology of installing data mapping unit 700, there is cluster (clustering) technology of self-elevating platform ART, vector quantization etc.In addition, the model used in diagnosis is not limited to above-mentioned clustering method, can also use the statistical model of the model, neural network etc. that make use of physically.
Be arranged in the diagnosis unit 800 in described diagnostic device 200, relative to the input of measuring-signal 4, by referring to the model definition information 7 of pattern definition number according to storehouse 320 and the model information 9 of diagnostic model database 330, the running status of equipment 100 is diagnosed, and exports this diagnostic result 10.
Diagnosis unit 800 the diagnostic result 10 of the current running status relative to equipment 100 diagnosed out be sent to image display device 940 set in operational management room 900 via outside output interface 220, as image display information 11, and shown.Thus, the diagnostic result of the running status relative to equipment 100 is informed to the operating personnel being positioned at operational management room 900.
So, in the diagnostic device 200 of the equipment of the present embodiment, the state to operating personnel's announcement apparatus there occurs the situation of change.
In addition, the measuring-signal database 310 in diagnostic device 200, pattern definition number at random can be shown in operational management room 900 image display device 940 according to the diagnostic device information 50 of preserving in storehouse 320, diagnostic model database 330 is arranged at.In addition, these information by operating the external input device 910 be made up of keyboard 920 and mouse 930, revise by the external input signal 2 generated.
Next, the action of the diagnostic device of the equipment of the present embodiment is described.Below, use the process flow diagram of the elemental motion of the diagnostic device representing the equipment shown in Fig. 1 and Fig. 2 (a) that the action flow chart of diagnostic device 200 is described.
As shown in the process flow diagram of Fig. 2 (a), the elemental motion combination step 201,202,203 of diagnostic device 200 performs.
First, in step 201, the pattern of judgement diagnostic device 200 is data mapping pattern or diagnostic mode.Then, proceeding to step 202 when data mapping pattern, proceeding to step 203 when being diagnostic mode.
If make step 202 action, then model definition unit 400, data mapping unit 700 action.Its result, generation model definition information 5 and model information 8, the information be made is stored in pattern definition number respectively according in storehouse 320, diagnostic model database 330.Figure 10 (a) is used to describe the detailed of the action of data mapping pattern later.
In addition, if make step 203 action, then diagnosed by the running status of diagnosis unit 800 to equipment 100, by the image display information 11 comprising diagnostic result is sent to image display device 940, the running status of display device 100 on image display device 940.Figure 10 (b) is used to describe the detailed of the action of diagnostic mode later.
The timing making data mapping pattern and diagnostic mode action of diagnostic device 200 at random can be specified by operating personnel.Below, use Fig. 2 (b) ~ (d) that the various embodiments making the timing of data mapping pattern and diagnostic mode action are described respectively.
In the embodiment shown in Fig. 2 (b), make data mapping pattern and diagnostic mode action in the sampling period of each measuring-signal, diagnose thus.
By all upgrading diagnostic model whenever obtaining measuring-signal, the diagnosis of up-to-date model always can be used.
But, when being used in the data volume in data mapping and being more, owing to needing the time in data mapping, therefore there is the possibility that cannot terminate to calculate within the sampling period.
Under these circumstances, the embodiment as shown in Fig. 2 (c), can also make normal condition data mapping Modal action during the setting of each regulation, each sampling period only makes diagnostic mode action, diagnoses thus.In the method for the embodiment shown in Fig. 2 (b) and Fig. 2 (c), just perform diagnostic mode whenever the sampling period, can the state of diagnostic device online.
In addition, the embodiment as shown in Fig. 2 (d), by being inputted the external input signal 2 constructed for implementation model, diagnose to diagnostic device 200 by operating personnel, data mapping pattern and diagnostic mode action can be made in arbitrary timing.That is, the running status of various condition to equipment 100 can be changed to diagnose.
Next, use Fig. 3, Fig. 4 that the function of classifying to the measuring-signal 4 of equipment 100 possessed in the data mapping unit 700 of diagnostic device 200 of the diagnostic device of the equipment forming the present embodiment and diagnosis unit 800 is described.
In the diagnostic device of the equipment of the present embodiment, carry out using the situation of self-elevating platform ART (AdaptiveResonanceTheoty:ART) in Data classification function describing.In addition, as Data classification function, other the clustering method such as vector quantization can be used.
As shown in Fig. 3 (a), Data classification function is made up of data pre-processing device 710 and ART module 720.Service data is transformed to the input data of ART module 720 by data pre-processing device 710.
Below, their order (operation) that described data pre-processing device 710 and ART module 720 carry out is described.
First, in data pre-processing device 710, be used in pattern definition number and by each measure the item, data be normalized according to the information of the normalizing condition of preservation in storehouse 320.The data of the complement CNxi (n) (=1-Nxi (n)) comprising the data Nxi (n) after being normalized measuring-signal and the data after normalization are set to input data Ii (n).This input data Ii (n) is imported in ART module 720.
In ART module 720, the measuring-signal 4 of the equipment 100 as input data is categorized as multiple classification.
ART module 720 possesses F0 layer 721, F1 layer 722, F2 layer 723, storer 724 and chooser system 725, and they are bonded to each other.F1 layer 722 and F2 layer 723 combine via weight coefficient.Weight coefficient represents the prototype (prototype) to the classification that input data are classified.At this, prototype is the type of other typical value of representation class.
Next, the algorithm of ART module 720 is described.
Input data are inputed to the summary of the algorithm of the situation of ART module 720, as following process 1 ~ process 5.
Process 1: by F0 layer 721 by input vector normalization, except denoising.
Process 2: by the input data and weight coefficient that are input to F1 layer 722 being compared, select the candidate of the classification be applicable to.
Process 3: by evaluating the rationality of the classification selected by chooser system 725 with the ratio of parameter ρ.If be judged as rationally, then will input Data classification for this classification, proceed to process 4.On the other hand, rationally if be not judged as, then reset this classification, from other classification, select the candidate of the classification be applicable to (repeatedly processing 2).As comparatively large in made the value of parameter ρ, then the classification of classification is just careful, if make the value of parameter ρ less, then classifies and just becomes rough.This parameter ρ is called warning (vigilance) parameter.
Process 4: if reset whole known classifications in process 2, be then judged as that input data belong to new classification, generate the new weight coefficient of the prototype representing new classification.
Process 5: if be classification J by input Data classification, then corresponding with classification J weight coefficient WJ (new: new) uses weight coefficient WJ (old: old) in the past and input data p (or by inputting the data of data fork), upgrades according to following formula (1).
[several 1]
WJ (new)=Kwp+ (1-Kw) WJ (old) ... formula (1)
At this, Kw is Study rate parameter (0 < Kw < 1), is the value determining input vector to be reflected in the degree in new weight coefficient.
In addition, formula (1) and formula described later (2) each arithmetic expression to formula (12) is embedded in described ART module 720.
The feature of the data classification algorithm of ART module 720 is above-mentioned process 4.
In process 4, when the input data that the pattern (pattern) that have input from carried out when learning is different, record new pattern while the pattern recorded can not be changed.Therefore, the pattern of the study of recording over records new pattern.
So, if give the service data of giving in advance as input data, then ART module 720 learns given pattern.Therefore, if input new input data to the ART module 720 completing study, then judge these input data are close to which pattern in the past by above-mentioned algorithm.In addition, if the pattern that the past does not live through, then new classification is categorized as.
Fig. 3 (b) is the block diagram of the formation representing F0 layer 721.In F0 layer 721, in each moment once again to input data I
ibe normalized, be made the normalization input vector u being input to F1 layer 721 and chooser system 725
i 0.
First, according to input data I
i, calculate w according to formula (2)
i 0, at this, a is constant.
[several 2]
W
i 0=I
i+ au
i 0formula (2)
Next, formula (3) is used to calculate W
i 0the x obtained after being normalized
i 0.At this, || || be the mark representing norm (norm).
[several 3]
Then, use formula (4), calculate from x
i 0in except the V after denoising
i 0.Wherein, θ is for the constant except denoising.The calculating of through type (4), because small value becomes 0, therefore, eliminates the noise of input data.
[several 4]
Finally, formula (5) is used to ask for normalization input vector u
i 0.U
i 0it is the input of F1 layer.
[several 5]
Fig. 3 (c) is the block diagram of the formation representing F1 layer 722.In F1 layer 722, the u that will ask in formula (5)
i 0kept as short-term storage, calculated the p of input in F2 layer 722
i.The calculating formula of F2 layer is shown in gathering formula (6) ~ formula (12).Wherein, a, b are constants, and f () is the function represented by formula (4), T
jcarry out at F2 layer 722 grade of fit that calculates.
" several 6]
W
i=u
i 0+ au
iformula (6)
[several 7]
[several 8]
V
i=f (x
i)+bf (q
i) ... formula (8)
[several 9]
[several 10]
[several 11]
Wherein,
[several 12]
Next, use Fig. 4 that possess in the data mapping unit 700 of the diagnostic device 200 of the diagnostic device of the equipment forming the present embodiment, construct model with the measuring-signal 4 of equipment 100 function is described.
First, use Fig. 4 (a) to carry out the embodiment of devices illustrated 100, describe the information be included in measuring-signal 4.Next, use Fig. 4 (b) and Fig. 4 (c) to describe appearance measuring-signal 4 being categorized into classification.
Fig. 4 (a) is the embodiment of indication equipment 100 and the block diagram of steam power plant.
In Fig. 4 (a), steam power plant 100 comprises gas turbo-generator 110, control device 120 and data sending device 130.Gas turbo-generator 110 comprises generator 111, compressor 112, burner 113 and turbine 114.
When generating electricity, the air sucked being compressed and form pressurized air by compressor 112, this pressurized air being sent into burner 113, burns with fuel mix.Using the gases at high pressure produced by burning that turbine 114 is rotated, being generated electricity by generator 111.
In control device 120, correspondingly control the output of gas turbo-generator 110 with electricity needs.In addition, control device 120 using the service data 102 measured by the sensor (not shown) being arranged at gas turbo-generator 110 as input data.Service data 102 is quantity of states of suction temperature, fuel input amount, turbine exhaust gas temperature, secondary speed, generator power output, turboshaft vibration etc., measures in each sampling period.In addition, the weather information of atmospheric temperature etc. is also measured.
In control device 120, these service datas 102 are used to calculate control signal 101 for controlling gas turbo-generator 110.In addition, in control device 120, also implement the process given the alarm when the value of service data 102 has departed from the scope preset.Alarm signal, as being " 1 " when service data 102 has departed from the scope preset, being the digital signal of " 0 " time in scope, processing thus.When alarm signal is " 1 ", with sound or picture display etc. by the context notification of alarm to operating personnel.
The measuring-signal 1 of the service data 102 comprised measured by control device 120 and the control signal 101 calculated by control device 120 and alarm signal is sent to diagnostic device 200 by signal data dispensing device 130.
Fig. 4 (b) illustrates the figure measuring-signal 1 obtained from equipment 100 being categorized into the result of classification.Transverse axis is the time, and the longitudinal axis is measuring-signal, class number.Fig. 4 (c) is the figure of the example measuring-signal 1 of equipment 100 being categorized into the classification results of classification.
Fig. 4 (c), as an example, shows 2 projects in measuring-signal, marks with the chart of two dimension.In addition, the longitudinal axis and transverse axis have carried out normalization to represent to the measuring-signal of each project.
Measuring-signal is split into multiple classification 1000 (circle shown in Fig. 4 (c)) by the ART module 720 of Fig. 3 (a).A circle is equivalent to 1 classification.
In the present embodiment, measuring-signal is classified as 4 classifications.Class number 1 represents the group that the value of project A value that is comparatively large, project B is less, class number 2 represents the group that the value of project A, project B is all less, class number 3 represents the group that the value of project A is less, the value of project B is larger, and class number 4 represents the group that the value of project A, project B is all larger.
As shown in Fig. 4 (b), the data between the normal epoch before diagnosis starts are classified as classification 1 ~ 3.The data of the first half after diagnosis starts are classified as classification 2, are the classifications identical with between normal epoch.In this case, because the tendency of data is identical with between normal epoch, be therefore diagnosed as normal.On the other hand, the later half data after diagnosis starts are classified as classification 4, are the classifications different between normal epoch.Because the tendency of data is different, therefore there is the state change of equipment, abnormal possibility occurs.In this case, in diagnostic device 200 of the present invention, there are existing the operating personnel that abnormal this situation of possibility is shown to equipment in image display device 940, informing operating personnel.
In addition, in the present embodiment, describe the example measuring-signal of 2 projects being categorized as classification, but also can, to measuring-signals more than 3 projects, use multidimensional coordinate to be categorized as classification.
Fig. 5 is the figure of the relation of originate mode in the generating set shown in key diagram 4 (a) and process values, shows the change in time of output order value and turbo-machine temperature.
Representatively originate mode, has hot start and cold start.The situation of carrying out restart under the state by turbine and compressor being heat is called hot start.In addition, will stop through the long time, the situation of carrying out restart under the state that turbine, compressor cool down is called cold start.As shown in Fig. 5 (b), if originate mode is different, then the variation range starting the temperature in turbo-machine temperature when starting, load change is also different.
In existing diagnostic device, no matter how service condition all constructs 1 diagnostic model, therefore needs normalization scope to determine as comprising data.That is, such as, the higher limit of normalization scope is set to 1010, lower limit is set to 1011.
Such as, when the low load of hot start, compared with the scope 1020 changed with process values, normalization scope 1022 broadens.
If normalization wider range compared with the variation range of data, then the change of the value after normalization diminishes.Therefore, the change of data when extremely occurring cannot be caught, sometimes also can not produce new classification when abnormal generation.This becomes the reason failed to report.
The model definition unit 400 that the application of the invention possesses, can come suitably to determine normalization scope with service condition matchingly.By this function, can suppress to fail to report, and improve diagnostic accuracy.Below, concrete method is described.
Fig. 6 is the process flow diagram that the inscape of model definition unit 400 and the action of service condition detection unit 500 are described.As shown in Figure 6, this algorithm combination step 510,520,530,540,550 performs.
First, in step 510, by measuring-signal 4 savings Data Segmentation be constant load run in, in starting operation, out of service in, load change operating each period.
When output does not change, when the rate of change of the measuring-signal namely exported is less, be set in constant load operation.In addition, according to be included in the measuring-signal of generating set for distinguish in starting, stop in, signal in load change, to be divided in starting operation, out of service in, load change operating period.
In the operating situation of constant load, proceed to step 520, in starting operation, proceed to step 530, proceed to step 540 in out of service, under load changes operating situation, proceed to step 550.
In step 520, the information relevant to load-strap is extracted.Such as, be set to low output by 0 ~ 50% of specified output, be set to high output by 50 ~ 100%, according to output, load-strap classified.In step 530, extract to the kind of originate mode, start in the relevant information of loading rate.As originate mode, there are hot start pattern, cold start mode etc.In step 540, extract when starting to stop mode, stopping action and export relevant information.As stop mode, there is the pattern of arrestment in usually using continuously, carry out the pattern etc. of emergency cut-off when abnormal generation.In step 550, extract when changing start to loading rate, load export, load exports relevant information at the end of changing.
In addition, in the present embodiment, in order to distinguish in step 520,530,540,550 constant load run in, in starting operation, out of service in, load changes operating state, extract foregoing information, but this quantity of information can also be increased.Such as, as long as the information that secondary speed, speed-raising rate, the kind being supplied to the fuel of equipment, atmospheric temperature etc. process the measuring-signal of equipment 100 and obtain, then also can be appended in the information extracted in step 520,530,540,550.
As shown in Fig. 6 (b), the diagnostic model 710 employing identical measure the item is constructed with the set being categorized as the model of sub-diagnostic model 720 based on the information extracted at Fig. 6 (a).Define different normalizing conditions to diagnose to every sub-diagnostic model.
Below, the clear and decided embodiment of determining the normalizing condition determination section 600 of normalizing condition of Fig. 7 ~ Fig. 9 is used.The information of the normalizing condition using normalizing condition determination section 600 to determine, by data pre-processing device 710 by measuring-signal normalization.If the data items number of measuring-signal xi is N number of, the n-th measuring-signal is x (n).Data Nxi (n) after normalization following formula (13) represents.Wherein, Nmin (n) is normalized lower limit, and Nmax (n) is normalized higher limit.
[several 13]
Nxi (n)=(xi (n)-Nmin (n))/(Nmax (n)-Nmin (n)) ... formula (13)
In normalizing condition determination section 600, determine Nmin (n), the Nmax (n) shown in (13) formula.
Fig. 7 is the key diagram of the 1st embodiment of normalizing condition determination section 600.
Fig. 7 (a) is the process flow diagram of the 1st embodiment of normalizing condition determination section 600.As shown in Fig. 7 (a), this algorithm combination step 611,612,613 performs.
First, in step 611, carry out parted pattern by each service condition to construct and use data.In step 612, the information of measuring-signal amplitude of variation is extracted by each service condition.
In step 613, determine the higher limit Nmax1 (n) of normalization scope, the lower limit Nmin1 (n) of normalization scope.
In the operating situation of constant load, use (14), (15) formula.At this, Dmax1 (n) is the maximal value of measuring-signal, and Dmin1 (n) is minimum value.α, β are constants.
[several 14]
Nmax1 (n)=Dmax1 (n) × (1+ α) ... formula (14)
[several 15]
Nmin1 (n)=Dmin1 (n) × (1-β) ... formula (15)
In load change, in starting, in stopping, coupling output order value, decides normalization scope like that such as formula (16), (17).
[several 16]
Nmax2=MW*a+b ... formula (16)
[several 17]
Nmin2=MW*c+d ... formula (17)
At this, MW is output order value, and a, c are slopes when carrying out first-order approximation to the measuring-signal in load change.In addition, b, d are the values carrying out in formula (18), (19) calculating.
[several 18]
B=f+Dmax2 × 1.2 ... formula (18)
[several 19]
D=f-Dmin2 × 1.2 ... formula (19)
At this, f cuts square when carrying out first-order approximation to the measuring-signal in load change, and Dmax2 is the maximal value of measuring-signal being carried out to the deviation between the straight line of first-order approximation and measuring-signal, and Dmin2 is the minimum value of deviation.
Fig. 7 (b) is the time dependent figure of output when load being described from from low output to high exporting change and measuring-signal.Until moment T0a is low output, be load change from moment T0a to moment T0b, be high output after moment T0b.
By making the process flow diagram action shown in Fig. 7 (a), the higher limit of normalization scope during low output is determined to be 1030, lower limit is determined to be 1040, the higher limit of the normalization scope in load being changed determines to be 1050, lower limit is determined to be 1060, the higher limit of normalization scope when being exported by height determines to be 1070, and lower limit is determined to be 1080.So, normalization scope of the present invention and service condition change matchingly.
Fig. 8 is the key diagram of the 2nd embodiment of normalizing condition determination section 600, and as shown in Fig. 8 (a), this algorithm combination step 631,632,633 performs.
In step 631, extract the data of load changing pattern.In step 632, the meaningless time is estimated by each data items.Using process signal as inputoutput data, estimate the meaningless time according to pulse (impulse) response.As concrete computing method, list of references " the superior system mark for controlling " (Tokyo motor university press office) (" the upper Grade シ ス テ system of the imperial め of system is with fixed " (East Jing Electricity Machine university press office) can be enumerated) in described in method.In step 633, after the data of amount being displaced the meaningless time estimated in step 632, decide normalization scope by each data items.
Use Fig. 8 (b) that this appearance is described.Start the moment T1a increased compared with output order value, the moment that measuring-signal starts to change is slower.This time is exactly the meaningless time.The meaningless time of data items A is (T2a-T1a), and the meaningless time of data items B is (T3a-T1a).
Make using the process flow diagram of Fig. 8 (a) after measuring-signal deviate from the amount of meaningless time, to use the process flow diagram of Fig. 7 to decide normalization scope.Thus, the normalization scope deciding measuring-signal can be coordinated with output order value.
Fig. 9 is the key diagram of the 3rd embodiment of normalizing condition determination section 600 in the diagnostic device of the generating set shown in Fig. 1.
By the normalization expanded range near the maximal value of measuring-signal, minimum value, thus be easy to the change of the measuring-signal detected near maximal value, minimum value.
In addition, the method for normalizing of normalizing condition determination section 600 is not limited to above-mentioned content, as long as decide normalizing condition by each service condition.Such as, the scope that data estimator value changes is carried out, using this scope as normalization scope according to the design information of equipment, the specification of measuring appliance.
Figure 10 is the process flow diagram of the pattern of the diagnostic device 200 of the generating set shown in key diagram 1.Figure 10 (a) is the action flow chart of the data mapping pattern in Fig. 2, and Figure 10 (b) is the action flow chart of diagnostic mode.
As shown in Figure 10 (a), data mapping mode combinations step 1200,1210,1220,1230 performs.In step 1200, be extracted in the data during using data mapping from measuring-signal database 310.Operating personnel during this period by equipment 100 at random set.Next, service condition detection unit 500 action is made in step 1210.Process flow diagram shown in Fig. 6 (a) carries out action, by the Data Segmentation during using in data mapping for constant load run in, in starting operation, out of service in, the operating each operational mode of load change, and then make step 520,530,540,550 actions, split the data of each pattern by each feature.Next, normalizing condition determination section 600 action is made in step 1220.Make the process flow diagram illustrated in Fig. 7 (a), Fig. 8 (a) carry out action, decide normalizing condition by each group of splitting in step 1210.Make step 1210,1220 actions and the model definition information 5 obtained is kept at pattern definition number according in storehouse 320.Finally, in step 1230, make data mapping unit 700 action.Each group that splits in step 1210 is defined as sub-diagnostic model 720 (with reference to Fig. 6 (b)), uses the normalizing condition defined by every sub-diagnostic model to process measuring-signal, use the ART described in Fig. 3 to construct diagnostic model.
As shown in Figure 10 (b), diagnostic mode combination step 1300,1310,1320 performs.In step 1300, from measuring-signal database 310, the data during carrying out diagnosing are extracted.Next, in step 1310, the data between diagnostic period are processed with service condition detection unit 500.The sub-diagnostic model that service condition is consistent is extracted from the multiple sub-diagnostic model constructed data mapping pattern.Finally, diagnosis unit 800 action is made in step 1320.In diagnosis unit 800, from pattern definition number according to being extracted in the sub-diagnostic model information extracted in step 1330 storehouse 320.The class number of sub-diagnostic model and the class number of classifying to the data between diagnostic period with ART and obtaining are compared.When being the classification identical with when making data mapping Modal action, being diagnosed as normal, when creating different class number, being diagnosed as exception.Diagnostic result 10 is outputted to outside output interface 220.
Figure 11 is the figure of the form that the data be kept in database of the present invention are described.
As shown in Figure 11 (a), in measuring-signal database 310, the value of service data and the measuring-signal 1 (in the drawings, describing data items A, B, C) that equipment 100 is measured is preserved by each sampling period (moment of the longitudinal axis).
In display frame 311, by using the scrolling bar 312 and 313 of energy cross shifting, the data of energy roll display wide region.
In pattern definition number according in storehouse 320, as shown in Figure 11 (b), the information of service condition and normalization scope is set up correspondence to preserve.
In diagnostic model database, as shown in Figure 11 (c), the relation of retention class numbering and weight coefficient.At this, weight coefficient is the centre coordinate of classification.
Figure 12 is the figure of the effect of the diagnostic device of the generating set shown in key diagram 1.
Start load change at moment T4, terminate load change at moment T5.After load change terminates, there occurs exception at moment T6.
Output order value is illustrated together with the relation of the measuring-signal of class number, temperature with the relation of turbine temperature by Figure 12.With abnormal generation, temperature rises, but due to be classification 4 scope in, be therefore classified as in the classification identical with normal condition.In existing mode, expand normalization scope, cannot catch and rise with the abnormal temperature occurred, can not exception be detected.
In the present invention, mate service condition and switch diagnostic model.In the present embodiment, according to until moment T4 service condition 1 diagnostic model (constant load run in, low output model) diagnose, diagnose with service condition 2 diagnostic model (in load change operation) between moment T4 ~ T5, switch by the mode that service condition 3 diagnostic model (constant load runs middle and high output model) carries out diagnosing after moment T5.
In service condition 3 diagnostic model, normal condition is categorized as class number 1 ~ 5.Be classified as the classification identical with normal condition before abnormal generation, but the data after abnormal generation are classified as the new classification of class number 6, can detect exception.That is, owing to classifying to state more meticulously than prior art, therefore can suppress to detect abnormal failing to report.
Figure 13 is the figure of picture shown in the image display device 940 in the diagnostic device 200 of the generating set shown in key diagram 1.
Operate cursor 951 with mouse 930, thus at random can adjust the higher limit (954,956,958) of normalization scope, the lower limit (955,957,959) of normalization scope, the border moment (952,953) of service condition.In order to be reflected in model by the result of adjustment, the picture of Figure 13 is clicked executive button 960.By this operation, change the service condition of the pattern definition number shown in Figure 10 (b) according to storehouse and the information of normalization scope, adjustment result can be reflected in data mapping and diagnostic action.
In addition, the present invention is not limited to the above embodiments, also comprises various variation.Such as, for above-described embodiment has carried out detailed record for ease of the present invention being described with understanding, but the whole formations possessing explanation are not limited to.
In addition, above-mentioned each formation, function, handling part, processing unit etc., also can by integrated circuit to design etc. with hardware realize wherein part or all.In addition, above-mentioned each formation, function etc. also can realize the program of each function and the software performed realizes by explaining for the treatment of device.Realize the program of each function, form, file, measuring-signal, the information that calculates information etc. can be placed in the memory storage of storer or hard disk etc. or the storage medium of IC-card, SD card, DVD etc.Therefore, each process, each formation can realize as process component, program module.
In addition, explanation is thought and needs and show information wire, but goods might not illustrate whole control lines or information wire.In fact, also can think that most formation is all connected to each other.
According to the present embodiment, can obtain detecting the diagnostic device of the generating set of the exception of generating set and the diagnostic method of equipment accurately.
Utilizability in industry
The present invention can be widely used in various equipment etc. as the diagnostic method of the diagnostic device of equipment and equipment.
Claims (14)
1. a diagnostic device for generating set, carry out the running status of diagnostic device based on the measuring-signal obtained from generating set measuring state amount, and be shown in image display device by diagnostic result, the feature of the diagnostic device of this generating set is, possesses:
Data mapping unit, it is used in the quantity of state measuring generating set in the diagnostic device of generating set and the measuring-signal obtained, and constructs the model used in diagnosis;
Model definition unit, the method for normalizing of service condition and the measuring-signal diagnosed is carried out in its definition by described model; With
Diagnosis unit, it uses the model constructed by described data mapping unit to diagnose the running status of generating set,
Possess in described model definition unit:
Service condition detection unit, it judges the service condition of generating set; With
Normalizing condition determination section, it decides the normalizing condition of measuring-signal by each service condition judged by service condition detection unit,
In described diagnosis unit, mate with service condition and switch diagnostic model, thus diagnose.
2. the diagnostic device of generating set according to claim 1, is characterized in that,
Described service condition detection unit possesses:
The service condition of generating set is categorized as constant load run in, in starting operation, out of service in, load change any one arithmetic unit operating;
The arithmetic unit of load-strap is extracted under service condition is the operating situation of constant load;
The kind of originate mode and the arithmetic unit of loading rate is extracted when service condition is in starting operation;
The arithmetic unit that the kind of stop mode and stopping action starting to export is extracted when in out of service; With
Extract loading rate when in load change runs, the output of load change when starting, the arithmetic unit of output at the end of load change.
3. the diagnostic device of generating set according to claim 1, is characterized in that,
In described normalizing condition determination section, possess:
Carry out parted pattern by each service condition judged by described service condition detection unit and construct the first arithmetic unit using data;
Extract the second arithmetic unit of the information of the data variation amplitude of each service condition; With
Information based on described data variation amplitude decides the 3rd arithmetic unit of normalization scope.
4. the diagnostic device of generating set according to claim 3, is characterized in that,
The diagnostic device of described generating set possesses the arithmetic unit normalization scope of the measuring-signal in load change being set to the function of output order value.
5. the diagnostic device of generating set according to claim 3, is characterized in that,
In described normalizing condition determination section, make for load change in measuring-signal to estimate the meaningless time arithmetic unit and make measuring-signal be shifted the estimated meaningless time amount arithmetic unit action after, make the first arithmetic unit according to claim 3, the second arithmetic unit and the 3rd arithmetic unit action.
6. the diagnostic device of the generating set according to any one of claim 3 ~ 5, is characterized in that,
In described normalizing condition determination section, possesses the arithmetic unit of the normalization scope expanded near the maximal value, minimum value of measuring-signal.
7. the diagnostic device of generating set according to claim 1, is characterized in that,
The diagnostic device of described generating set possesses image display device, the moment that this image display device makes the time dependent trend chart of expression measuring-signal, switch the service condition judged by described service condition detection unit and the normalization overlapping ranges determined by described normalizing condition determination section get up to show, and at random changing the moment and described normalization scope that switch described service condition.
8. the diagnostic method of a generating set, the running status of diagnostic device is carried out based on the measuring-signal obtained from generating set measuring state amount, and diagnostic result is shown in image display device, the feature of the diagnostic method of this generating set is to possess following steps:
Judge the service condition determination step of the service condition of generating set;
The normalizing condition deciding step of the normalizing condition of measuring-signal is decided by each service condition judged by service condition detection unit;
Be used in the quantity of state measuring generating set in the diagnostic device of generating set and the measuring-signal obtained, construct the step of the model used in diagnosis;
The step of the method for normalizing of service condition and the measuring-signal diagnosed is carried out in definition by described model; With
Mate with service condition and switch diagnostic model, thus carry out the step diagnosed.
9. the diagnostic method of generating set according to claim 8, is characterized in that,
Judge that the step of the service condition of described generating set possesses:
The service condition of generating set is categorized as constant load run in, in starting operation, out of service in, load change any one step operating;
The step of load-strap is extracted under service condition is the operating situation of constant load;
The kind of originate mode and the step of loading rate is extracted when service condition is in starting operation;
The step that the kind of stop mode and stopping action starting to export is extracted when in out of service; With
Extract loading rate when in load change runs, the output of load change when starting, the step of output at the end of load change.
10. the diagnostic method of generating set according to claim 8, is characterized in that,
In described normalizing condition deciding step, possess:
Carry out parted pattern by each service condition judged by described service condition detection unit and construct the first step using data;
Extract the second step of the information of the data variation amplitude of each service condition; With
Information based on described data variation amplitude decides the third step of normalization scope.
The diagnostic method of 11. generating sets according to claim 10, is characterized in that,
The normalization scope possessing the measuring-signal in load being changed is set to the step of the function of output order value.
The diagnostic method of 12. generating sets according to claim 10, is characterized in that,
In described normalizing condition deciding step, perform the step of estimating the meaningless time for the measuring-signal in load change and make measuring-signal be shifted the estimated meaningless time amount step after, enforcement of rights requires first step, second step and third step described in 10.
The diagnostic method of 13. generating sets according to any one of claim 10 ~ 12, is characterized in that,
In described normalizing condition deciding step, possesses the step of the normalization scope expanded near the maximal value, minimum value of measuring-signal.
The diagnostic method of 14. generating sets according to claim 8, is characterized in that,
The moment make the time dependent trend chart of expression measuring-signal, switching to the service condition judged by described service condition detection unit and the normalization overlapping ranges determined by normalizing condition determination section get up to show, and at random can change the moment and described normalization scope that switch described service condition.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2011-197837 | 2011-09-12 | ||
| JP2011197837A JP5337853B2 (en) | 2011-09-12 | 2011-09-12 | Power plant diagnostic device and power plant diagnostic method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102999038A CN102999038A (en) | 2013-03-27 |
| CN102999038B true CN102999038B (en) | 2015-12-02 |
Family
ID=47927732
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210288454.4A Active CN102999038B (en) | 2011-09-12 | 2012-08-07 | The diagnostic device of generating set and the diagnostic method of generating set |
Country Status (2)
| Country | Link |
|---|---|
| JP (1) | JP5337853B2 (en) |
| CN (1) | CN102999038B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6116466B2 (en) * | 2013-11-28 | 2017-04-19 | 株式会社日立製作所 | Plant diagnostic apparatus and diagnostic method |
| EP3128385B1 (en) | 2014-04-01 | 2019-04-10 | TLV Co., Ltd. | Risk evaluation system for process system, risk evaluation program, and risk evaluation method |
| WO2016117021A1 (en) | 2015-01-20 | 2016-07-28 | 株式会社日立製作所 | Machine diagnosis device and machine diagnosis method |
| JP6438124B2 (en) * | 2015-04-20 | 2018-12-12 | 株式会社日立製作所 | Operation management system and operation management method |
| JP6606003B2 (en) * | 2016-03-31 | 2019-11-13 | 株式会社東芝 | Process diagnostic apparatus, process diagnostic method, and computer program |
| JP6802025B2 (en) * | 2016-09-30 | 2020-12-16 | メタウォーター株式会社 | Valve operation monitoring method and valve operation abnormality sign detection method |
| JP6772963B2 (en) * | 2017-06-05 | 2020-10-21 | トヨタ自動車株式会社 | Abnormality diagnosis device and abnormality diagnosis method |
| JP6830414B2 (en) * | 2017-06-28 | 2021-02-17 | 株式会社日立製作所 | Diagnostic device and diagnostic method |
| JP7138441B2 (en) * | 2018-01-11 | 2022-09-16 | 三菱電機株式会社 | Soundness evaluation device and power generation system |
| JP7097263B2 (en) * | 2018-08-28 | 2022-07-07 | 株式会社日立製作所 | Plant data analysis system |
| JP7408911B2 (en) * | 2018-10-26 | 2024-01-09 | 富士電機株式会社 | Anomaly detection device and anomaly detection method |
| JP7175752B2 (en) * | 2018-12-27 | 2022-11-21 | 株式会社クボタ | Diagnostic method for mechanical equipment and diagnostic device for mechanical equipment |
| WO2020152741A1 (en) * | 2019-01-21 | 2020-07-30 | 三菱電機株式会社 | Abnormality cause estimation device, abnormality cause estimation method, and program |
| WO2020166468A1 (en) * | 2019-02-15 | 2020-08-20 | パナソニックＩｐマネジメント株式会社 | Playback system, recording system, playback method, program for playback, recording method, and program for recording |
| JP7239354B2 (en) * | 2019-03-12 | 2023-03-14 | 株式会社日立製作所 | data classifier |
| JP2021096492A (en) * | 2019-12-13 | 2021-06-24 | 三菱電機株式会社 | Diagnostic device, power generation facility, data creation method, and diagnostic program |
| JP7222943B2 (en) * | 2020-02-26 | 2023-02-15 | 株式会社日立製作所 | Operation improvement support system, operation improvement support method, and operation improvement support program |
| JP2021181145A (en) * | 2020-05-20 | 2021-11-25 | オムロン株式会社 | Screw tightening failure determination device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129013A (en) * | 2011-01-21 | 2011-07-20 | 昆明理工大学 | Distribution network fault location method utilizing natural frequency and artificial neural network |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2809860B2 (en) * | 1990-10-18 | 1998-10-15 | 株式会社東芝 | Plant process input value generator |
| JPH11161327A (en) * | 1997-11-28 | 1999-06-18 | Mitsubishi Chemical Corp | Method and device for diagnosing abnormality of process |
| JP4313734B2 (en) * | 2004-07-16 | 2009-08-12 | ヤンマー株式会社 | Monitoring and diagnosis system |
| JP2010244149A (en) * | 2009-04-02 | 2010-10-28 | Hitachi Ltd | Data classification method, data classification device, diagnostic method and diagnostic device |
| JP5431235B2 (en) * | 2009-08-28 | 2014-03-05 | 株式会社日立製作所 | Equipment condition monitoring method and apparatus |
-
2011
- 2011-09-12 JP JP2011197837A patent/JP5337853B2/en active Active
-
2012
- 2012-08-07 CN CN201210288454.4A patent/CN102999038B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102129013A (en) * | 2011-01-21 | 2011-07-20 | 昆明理工大学 | Distribution network fault location method utilizing natural frequency and artificial neural network |
Non-Patent Citations (1)
| Title |
|---|
| "基于LM算法建立风电机组神经网络故障预警诊断模型";巨林仓 等;《热力发电》;20101231;第39卷(第12期);第44-49页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| JP5337853B2 (en) | 2013-11-06 |
| CN102999038A (en) | 2013-03-27 |
| JP2013061695A (en) | 2013-04-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102999038B (en) | The diagnostic device of generating set and the diagnostic method of generating set | |
| CN102870057B (en) | Plant diagnosis device, diagnosis method, and diagnosis program | |
| WO2012073289A1 (en) | Plant diagnostic device and plant diagnostic method | |
| US7181365B2 (en) | Diagnostic data detection and control | |
| US7930122B2 (en) | Evaluating anomaly for one-class classifiers in machine condition monitoring | |
| JP5199478B2 (en) | Plant diagnostic equipment | |
| US11061390B2 (en) | System fault isolation and ambiguity resolution | |
| JP5292477B2 (en) | Diagnostic device and diagnostic method | |
| US20180231394A1 (en) | Gas turbine sensor failure detection utilizing a sparse coding methodology | |
| CN104756029B (en) | A kind of system of the parts group of monitoring device | |
| CN107710089B (en) | Plant equipment diagnosis device and plant equipment diagnosis method | |
| JP2011243118A (en) | Monitoring diagnosis device and monitoring diagnosis method | |
| JP6830414B2 (en) | Diagnostic device and diagnostic method | |
| KR102383675B1 (en) | Anomaly detection system of time-series data | |
| JP6685124B2 (en) | Diagnostic device and diagnostic method | |
| King et al. | Probabilistic approach to the condition monitoring of aerospace engines | |
| JP2890815B2 (en) | Plant abnormality diagnosis device | |
| KR102486462B1 (en) | Method and Apparatus for Fault Detection Using Pattern Learning According to Degradation | |
| US20230046709A1 (en) | Prediction apparatus, prediction method, and program | |
| Zhang et al. | An Enhanced Joint Indicator for Starter Failure Diagnostics in Auxiliary Power Unit | |
| Saadawia et al. | SVM-based fault diagnosis system for materials change detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |