CN107710089A - Shop equipment diagnostic device and shop equipment diagnostic method - Google Patents
Shop equipment diagnostic device and shop equipment diagnostic method Download PDFInfo
- Publication number
- CN107710089A CN107710089A CN201680036343.0A CN201680036343A CN107710089A CN 107710089 A CN107710089 A CN 107710089A CN 201680036343 A CN201680036343 A CN 201680036343A CN 107710089 A CN107710089 A CN 107710089A
- Authority
- CN
- China
- Prior art keywords
- shop equipment
- diagnosis
- unit
- abnormal state
- diagnostic device
- 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
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
When detecting abnormal, whether should be handled by the micro-judgment of shop equipment operation abnormal, but preferably based on the risk in the case of ignoring extremely (loss forecasting volume) is judged.In order to solve above-mentioned problem, the present invention provides a kind of shop equipment diagnostic device, possess multiple diagnosis units of the abnormality of diagnosis shop equipment, it is characterized in that possesses comprehensive diagnos unit, the comprehensive diagnos unit is according to the measurement signal data relevant with the state of the shop equipment and the device management information data relevant with past abnormal state, the degree of accuracy of the respective detection on the abnormal state of the multiple diagnosis unit is obtained, according to the degree of accuracy and the loss associated with abnormal state come assessment of loss amount forceasted.
Description
Technical field
The present invention relates to the shop equipment diagnostic device and shop equipment of the abnormal state of diagnosis shop equipment (Plant)
Diagnostic method.
Background technology
In the case of abnormal transient phenomenon or accident etc. occur in shop equipment, shop equipment diagnostic device according to
Measurement data from shop equipment detects the generation of the exception or accident.
Patent Document 1 discloses the adaptive resonance theory of one of application cluster (clustering) technology
The diagnostic device of (Adaptive Resonance Theory (ART)).Here, so-called ART is the data by multidimensional according to it
Similarity is categorized as the theory of different classifications.
In the above-described techniques, measurement data when will be normal first with ART is categorized as multiple classifications (normal category).So
Afterwards, current measurement data is input to ART and is classified as different classifications.It can not be classified into just in the measurement data
During normal classification, new classification (new category) is generated.New category means that the state of shop equipment is changed.Cause
This, judges abnormal generation using new category, is diagnosed as in the case where the incidence of new category exceedes threshold value different
Often.
【Prior art literature】
【Patent document】
Patent document 1:Japanese Unexamined Patent Publication 2005-165375 publications
The content of the invention
The invention problem to be solved
In clustering technique, there is the size (in ART be classification (category) size) for determining group (cluster)
Parameter.This parameter is referred to as resolution parameter.In general, when a certain data are categorized as into different groups, if coarse
The quantity of ground setting resolution ratio then group will tail off, and the quantity of group becomes more if resolution ratio is subtly set.
When using clustering technique in abnormity diagnosis, the amplitude of variation for the data trend that new category occurs is according to resolution ratio
Coarse and resolution precision and it is different.When new category occurs in the case of coarse in resolution ratio, data trend and normal phase
Than changing a lot, therefore the degree of accuracy height that machine is abnormal.On the other hand, in the case of resolution precision, it is possible to
The change of faint trend as measurement noise is detected, therefore the degree of accuracy of machine exception is relatively low.Thus, it is determined that group
Size pre-set parameter difference when, it is different that the abnormal degree of accuracy occurs in abnormality detection.
In general, abnormality detection performance is different if diagnostic method difference, so exception occurs in abnormality detection
Accuracy in detection is different.
In addition, in abnormality detection, stop shop equipment and carry out maintenance and repair to avoid mechanical disorder be it is appropriate,
But, it may occur that for maintenance and repair expense and stop shop equipment during opportunity loss.Therefore, if gently
Micro- failure, also continue to run with sometimes untill inspecting periodically.On the other hand, the result ignored extremely may be caused
Mechanical disorder, damage, loss are bigger when will be than maintenance and repair.
Present situation is when detecting abnormal, decides whether to enter exception by the experience of shop equipment operation
Row processing, but had better be judged based on the risk (loss forecasting volume) when ignoring exception.
Means for solving the problems
In order to solve above-mentioned problem, the present invention provides a kind of shop equipment diagnostic device, possesses multiple diagnosis shop equipments
Abnormal state diagnosis unit, the shop equipment diagnostic device is characterised by possessing comprehensive diagnos unit, the comprehensive diagnos
Unit is according to the measurement signal data relevant with the state of the shop equipment and the equipment relevant with past abnormal state
Management information data, the degree of accuracy of the respective detection on the abnormal state of the multiple diagnosis unit is obtained, according to institute
The degree of accuracy and the loss associated with abnormal state are stated to estimate loss forecasting volume.
The effect of invention
Loss forecasting volume is obtained in abnormality detection, and can be provided to judging whether at the exception to detecting
Manage useful information.
Brief description of the drawings
Fig. 1 is the block diagram for the diagnostic device for illustrating the 1st embodiment as the present invention.
Fig. 2 is the flow for the action for illustrating the comprehensive diagnos unit under the evaluation profile of diagnostic device and diagnostic mode
Figure.
Fig. 3 figures are the figures that explanation makes the timing of evaluation profile and diagnostic mode action.
Fig. 4 is the figure for the situation that explanation is stored in measurement signal database and the data in device management information database.
Fig. 5 is the figure of the situation for the data that explanation is stored in diagnostic result database.
Fig. 6 is the explanation figure of adaptive resonance theory.
Fig. 7 is the figure for showing measurement signal being categorized as the result of different classifications.
Fig. 8 be illustrate classification size with detection timing, the degree of accuracy, loss forecasting volume relation figure.
Fig. 9 is the testing result and the figure of the rheological parameters' change with time of loss forecasting volume for illustrating each diagnosis unit.
Figure 10 is the figure for the modification method for illustrating the degree of accuracy.
Figure 11 is the figure of the embodiment for the picture that explanation is shown in image display device.
Figure 12 is the figure of the embodiment for the picture that explanation is shown in image display device.
Figure 13 is the figure for illustrating Model Diagnosis.
Figure 14 is the figure of explanation effect as and with obtained from clustering technique diagnosis and Model Diagnosis.
Figure 15 is the figure of embodiment when illustrating for the diagnostic device of the present invention to be applied to steam power plant.
Description of reference numerals
1:Measurement signal;2:External input signal;3:Measurement signal;4:Device management information signal;5:Measurement signal;6:
Device management information signal;7:Diagnostic result database information;8:Diagnostic result database information;9:Measurement signal;10:Diagnosis
As a result;11:Comprehensive diagnos consequential signal;12:Comprehensive diagnos consequential signal;100:Shop equipment;200:Diagnostic device;210:Number
According to input interface;220:Data output interface;300:Measurement signal database;310:Device management information database;320:Examine
Disconnected result database;400:Comprehensive diagnos unit;500:Diagnosis unit;800:Image display device;900:External input device;
910:Keyboard;920:Mouse
Embodiment
Hereinafter, with reference to the accompanying drawings of the diagnostic device of suitable implementation of the invention.In addition, the description below is only to implement
Example, its objective are not intended for invention to be defined in following particular contents in itself.
Embodiment 1
Fig. 1 is the block diagram for the diagnostic device for illustrating the 1st embodiment as the present invention.Diagnostic device 200 is set with factory
Standby 100, image display device 800 and external input device 900 are connected, and shop equipment 100 is monitored, diagnosed.Separately
Outside, diagnostic device 200 be configured to be connected in a wired or wireless manner performed between each machine or device communication communication unit,
Computer and computer server (CPU:Central Processing Unit:CPU), memory and various
Database D B etc..In addition, external input device 900 is referred to by the instruction device of keyboard switch, mouse etc., touch panel and sound
Showing device etc. is formed, and image display device 800 is made up of liquid crystal display etc..
Diagnostic device 200 has the comprehensive diagnos unit 400 and diagnosis unit 500 as arithmetic unit.Diagnosis unit
500 there is multiple and its quantity can arbitrarily set.In addition, diagnostic device 200 has the measurement signal data as database
Storehouse 300, device management information database 310 and diagnostic result database 320.In addition, database is abbreviated as in Fig. 1
DB。
Preserved in measurement signal database 300, device management information database 310 and diagnostic result database 320
There is the information of electronization, and information is preserved in the form of being commonly referred to as e-file (electronic data).
In addition, diagnostic device 200 has outer input interface 210 and the outside output interface as the interface with outside
220。
Also, the various states that will be measured via outer input interface 210 as the running status for detecting shop equipment 100
Measurement signal 1 obtained from amount and utilization are arranged at the operation of the keyboard 910 and mouse 920 of external input device 900 and made
External input signal 2 be taken into diagnostic device 200.In addition, via outside output interface 220 by comprehensive diagnos consequential signal 12
It is output to image display device 800.
In the diagnostic device 200 shown in Fig. 1, the various of measurement shop equipment 100 are taken into via outer input interface 210
Detection signal 1 obtained from quantity of state.The measurement signal 3 for being taken into diagnostic device 200 is stored in measurement signal database
300.In addition, in shop equipment 100 caused fault message, safeguard the shop equipment management information of information etc. by using key
The operation of disk 910 and mouse 920 and caused external input signal 2 and be taken into diagnostic device 200.It is taken into diagnosis dress
The device management information signal 4 for putting 200 is saved in device management information database 310.
Diagnostic device 200 has evaluation profile and diagnostic mode both tupes.On evaluation profile and diagnosis mould
The flow and comprehensive diagnos unit 400 of formula, the action of diagnosis unit 500, quote Fig. 1 and Fig. 2 and illustrate.
Furthermore in the diagnostic device 200 of the present embodiment, although by comprehensive diagnos unit 400, diagnosis unit 500, measuring
Signals Data Base 300, device management information database 310 and diagnostic result database 320 are configured at the interior of diagnostic device 200
Portion, but it is also possible to which a part of device in these to be configured to the outside of diagnostic device 200, only communicate number between the devices
According to.
In addition, the information on being stored in the database for being arranged at diagnostic device 200, can show its all information
It is shown in image display device 100, the external input signal 1 that these information can be generated by operation external input device 900
Modify.
In the present embodiment, although forming external input device 900 but it is also possible to be for inputting language by keyboard and mouse
The means for entering data such as the microphone of sound, touch-screen.
In addition, as embodiments of the present invention, diagnostic method is can be used as certainly, offer makes diagnostic device 200 dynamic
The information providing services of information obtained from work are implemented.
Fig. 2 is the stream of the action of the comprehensive diagnos unit 400 under the evaluation profile and diagnostic mode for illustrating diagnostic device 200
Cheng Tu.
Fig. 2 (a) is the flow chart of evaluation profile.
First, in step 2000, comprehensive diagnos unit 400 extracts the regulation phase for being stored in measurement signal database 300
Between in measurement signal 5.
In step 2010, measurement signal 9 is sent to diagnosis unit 500 by comprehensive diagnos unit 400.Diagnosis unit 500
Processing measurement signal 9 simultaneously diagnoses the state of shop equipment 100, and diagnostic result 10 is sent into comprehensive diagnos unit 400.In synthesis
In diagnosis unit 400, collect the diagnostic result 10 received, diagnostic result database information 8 is sent to diagnostic result data
Storehouse 320 simultaneously preserves.
In step 2020, comprehensive diagnos unit 400 extracts the equipment pipe being stored in device management information database 310
Manage information signal 6.
In step 2030, by each diagnosis for the diagnostic result database information 7 being stored in diagnostic result database 320
The testing result of unit is compared with the device management information signal 6 extracted in step 2020, accuracy in computation and average
Leading time.Here, the degree of accuracy number of stoppages divided by detection number and try to achieve.In addition, it is average refer to leading time from
The time tried to achieve at the time of being detected in corresponding diagnosis unit is subtracted at the time of being detected according to threshold determination, is table
Show the time how far ahead of time detected.The degree of accuracy for each diagnosis unit obtained in step 2030 and average leading time preserve
In diagnostic result database 320.
In step 2040, comprehensive diagnos unit 400 extracts the diagnostic result number being stored in diagnostic result database 320
According to storehouse information 7, outside output interface 220 is sent to as comprehensive diagnos consequential signal 11.The quilt of comprehensive diagnos consequential signal 12
Image display device 800 is sent to, and is shown in image display device 800.
Fig. 2 (b) is the flow chart for the action for illustrating diagnostic mode.
In step 2100, comprehensive diagnos unit 400 extracts the phase diagnose being stored in measurement signal database
Between service data 5.
In step 2110, measurement signal 9 is sent to diagnosis unit 500 by comprehensive diagnos unit 400.Diagnosis unit 500
Processing measurement signal 9 simultaneously diagnoses the state of shop equipment 100, and diagnostic result 10 is sent into comprehensive diagnos unit 400.In synthesis
In diagnosis unit 400, collect the diagnostic result 10 received, diagnostic result database information 8 is sent to diagnostic result data
In storehouse 320 and preserve.
In step 2120, assessment whether there is abnormality detection, advances in the case where presence detects abnormal diagnosis unit
To step 2130, step 2160 is proceeded in the case of no.
In step 2130, comprehensive diagnos unit 400 extracts the diagnostic result number being stored in diagnostic result database 320
According to storehouse information 7, the information of the grasp degree of accuracy relevant with detecting abnormal diagnosis unit in step 2120.
In step 2140, comprehensive diagnos unit 400 extracts the equipment pipe being stored in device management information database 310
Information 6 is managed, grasps loss caused by failure.
In step 2150, comprehensive diagnos unit 400 is based on the degree of accuracy extracted in step 2130 and in step
The loss extracted in 2140 carrys out counting loss amount forceasted.On loss forecasting volume, have certainly the degree of accuracy is multiplied with loss,
Or a variety of obtain method with what predetermined parameter evaluate etc..
In step 2160, the testing result of each diagnosis unit is shown in image display device 800, is detected existing
During the diagnosis unit of exception, the loss forecasting volume that step 2150 calculates is additionally shown in image display device 800.
As described above, in the diagnostic device 200 of the present invention, when diagnosis unit 500 detects abnormal, display loss is pre-
Survey volume, thereby, it is possible to provide to judging whether to carry out processing useful information to the exception detected.
Fig. 3 is the figure that explanation makes the timing of evaluation profile and diagnostic mode action.
In method shown in (a) in Fig. 3, after the service data during accumulation is certain, evaluation profile is set once to be moved
Make, diagnostic mode is acted by certain cycle.
In method shown in (b) in Fig. 3, evaluation profile is acted at regular intervals, diagnosis is stored in renewal
After the degree of accuracy of result database 320, average leading time data, acted diagnostic mode.
In method shown in (c) in Fig. 3, when sending instruction from user, act evaluation profile.In arbitrary timing
Evaluation profile is performed, and updates the degree of accuracy and average leading time, acts diagnostic mode.
In addition, in addition to the timing stated in the present embodiment, make evaluation profile and what diagnostic mode was acted determines
When can also arbitrarily set.
Fig. 4 is the situation that explanation is stored in measurement signal database 300 and the data in device management information database 310
Figure.
As shown in Fig. 4 (a), in measurement signal database 300, preserved for each sampling period (time of the longitudinal axis)
Value as the measurement signal 1 (in figure, recording data items A, B, C) of the service data measured shop equipment 100.
By using in display picture 301 can cross shifting scroll box 302 and 303, can rollably show width
The data of scope.
As shown in Fig. 4 (b), defect content, countermeasure expense are preserved in device management information database 310, avoids event
Hinder the leading time needed, stop number of days caused by failure and because shop equipment stops and caused by opportunity loss volume etc. therefore
Hinder information.
In addition, as Fig. 4 (c) shown in, in device management information database 310 preserve maintenance content, safeguard needed for
The maintenance information of opportunity loss volume etc. caused by number of days, maintenance needed for expense, maintenance.
Fig. 5 is the figure of the situation for the data that explanation is stored in diagnostic result database 320.
As shown in Fig. 5 (a), in diagnostic result database 320, preserved for each sampling period (at the time of longitudinal axis)
There is the testing result (diagnosis unit A, B, C have been recorded in figure) of each diagnosis unit.
By using in display picture 311 can cross shifting scroll box 312 and 313, can rollably show width
The data of scope.
Preserve the testing result in each diagnosis unit in diagnostic result database 320, for example, as be 1 during unusual determination,
Normal is 0 diagnostic result is replaced into digital information like that to be preserved when judging.
As shown in Fig. 5 (b), in diagnostic result database, it is stored in evaluation profile for each diagnosis unit and falls into a trap
The degree of accuracy calculated and average leading time.
Fig. 6 describes the situation of the application adaptive resonance theory (ART) as the embodiment of diagnosis unit 500.In addition,
Other clustering methods such as vector quantization, SVMs can also be used.
As shown in Fig. 6 (a), data classification feature is made up of data prediction device 610 and ART modules 620.Data
Pretreatment unit 610 is converted to service data the input data of ART modules 620.
Hereinafter, the step of implementing to above-mentioned data prediction device 610 and ART modules 620 illustrates.
First, in data prediction device 610, data are normalized for each measure the item.It will contain and return
Data Nxi (n) and the complement CNxi (n) (=1-Nxi (n)) of normalized data that one change measurement signal obtains data conduct
Input data Ii (n).The input data Ii (n) is input into ART modules 620.
In ART modules 620, multiple classes are categorized as using as the measurement signal 10 of input data or operation signal 11
Not.
ART modules 620 have F0 layers 621, F1 layers 622, F2 layers 623, memory 624 and selection subsystem 625, they
It is combined with each other.F1 layers 622 and F2 layers 623 are combined by weight coefficient.Weight coefficient represents the classification that input data is classified
Initial form (prototype).Here, so-called initial form represents the typical value of classification.
Then, the algorithm of ART modules 620 is illustrated.
In the case where input data is input into ART modules 620, the summary of the algorithm turn into as following processing 1~
Processing 5.
Processing 1:Input vector is normalized by F0 layers 621, eliminates noise.
Processing 2:By being input to the input data of F1 layers 622 and the comparison of weight coefficient, the candidate of suitable class is selected.
Processing 3:The properness of the classification selected in subsystem 625 is selected is assessed by the ratio with parameter ρ.If
It is judged as appropriate, then input data is classified into the category and proceeds to processing 4.On the other hand, it is if it is determined that imappropriate, then
The category is reset and selects candidate's (reprocessing 2) of suitable classification from other classifications.If increasing the value of parameter ρ,
Becoming for the classification of classification is finer.That is, classification size diminishes.If on the contrary, reducing the value of parameter ρ, the classification of classification becomes
It is coarse.Classification becomes large-sized.This parameter ρ is called warning (vigilance) parameter.
Processing 4:When whole existing classifications in processing 2 is reset, it is judged as that input data belongs to new category, generates
Represent the new weight coefficient of a primitive type of new category.
Processing 5:When input data is classified as different classification J, weight coefficient WJ (new) corresponding with classification J is logical
(formula 1) is crossed to be carried out more with past weight coefficient WJ (old) and input data p (or by the derivative data of input data)
Newly.
(formula 1)
WJ (new)=Kwp+ (1-Kw) WJ (old)
Here, Kw is Study rate parameter (0<Kw<1), it is to determine input vector reflection to the degree of new weight coefficient
Value.
In addition, each arithmetic expression of formula 1 and formula 2 to formula 12 below is embedded in the ART modules 620.
The data classification algorithm of ART modules 620 is characterised by above-mentioned processing 4.
In processing 4, when being transfused to the input datas different from pattern during study, the mould of record can not be changed
Formula and record new pattern.Therefore, new pattern can be recorded while record study in the past is to pattern.
Thus, when providing the service data being provided previously by as input data, ART modules 620 learn what is provided
Pattern.So that, can be true by above-mentioned algorithm when new input data is input into the ART modules 620 for having learnt to finish
It is fixed whether to be approached with some past pattern.In addition, if being the pattern not lived through in the past, then it is classified as new category.
Fig. 6 (b) shows the block diagram of the structure of F0 layers 621.In F0 layers 621, at each moment again by input data Ii
Normalization, make the normalization input vector for being input to F1 layers 621 and selecting subsystem 625
First, according to formula 2, according to input data IiCalculateHere a is constant.
(formula 1)
Then, using the calculating pair of formula 3Obtained from normalizationHere, | | w0| | represent w0Norm.
(formula 2)
Then, using formula 4 calculate fromEliminate noiseWherein, θ is the constant for eliminating noise.According to public affairs
The calculating of formula 4, small value turn into 0, and therefore, the noise of input data is eliminated.
(formula 3)
Finally, normalization input vector is obtained using formula 5For the input of F1 layers.
(formula 4)
Fig. 6 (c) is the block diagram for the structure for showing F1 layers 622.In F1 layers 622, by what is obtained by formula 5Remain
Short-term storage, calculate the P for being input to F2 layers 722i.The calculation formula of F2 layers is collected and is shown as formula 6 to formula 12.Wherein,
A, b is constant, and f () is the function represented by formula 4, TjIt is the degree of fitting calculated by F2 layers 623.
(formula 5)
(formula 6)
(formula 7)
vi=f (xi)+bf(qi)
(formula 8)
(formula 9)
(formula 10)
Wherein, (formula 11)
Fig. 7 shows for measurement signal to be categorized as the figure of the result example of different classifications.
Fig. 7 (a) shows for the measurement signal 1 of shop equipment 100 to be categorized as one of the classification results of different classifications
The figure of example.
Fig. 7 (a) is used as an example, it is shown that two projects in measurement signal, and carried out with X-Y scheme
Mark.In addition, the longitudinal axis and transverse axis standardize and show each project survey signal.
Measurement signal is divided into (shown in Fig. 4 (c) by multiple classifications 630 by the ART modules 620 of Fig. 3 (a)
Circle).Equivalent to one classification of one circle.
In the present embodiment, measurement signal is categorized as four classifications.Classification number 1 is that project A value is big, project B value
Small group, classification number 2 are all small groups of project A, project B value, and classification number 3 is the group that project A value is small, project B value is big,
The all big group of 4 A of classification number, project B value.
Fig. 7 (b) is the result that the measurement signal 1 for illustrating to obtain from shop equipment 100 is categorized as different classifications
Figure.Transverse axis is the time, and the longitudinal axis is measurement signal, classification number.
As shown in Fig. 7 (b), the data for diagnosing the normal period before starting are classified as different classifications 1~3.Open
Begin monitoring after front half section data be classified as classification 2 and be and model data identical classification.In this case, due to
The trend of data is identical, so being judged as that state does not change.On the other hand, the data for starting the second half section of monitoring are classified as class
Other 4, it is classified as the classification different from model data.Because the trend of data is different, so being judged as the state of shop equipment
Changed.
Thus, in the diagnostic techniques of clustering technique is applied, there is the feature of detection data Long-term change trend.
Fig. 8 be illustrate the size of classification and detection timing, the degree of accuracy, loss forecasting volume relation figure.
As shown in Fig. 8 (a), by the parameter ρ for determining resolution ratio be set to it is big, reduce classification size when, even micro-
Small change can also detect.It can detect in advance.On the contrary, because detect the small change such as measurement noise, the degree of accuracy
Reduce.
On the other hand, when parameter ρ value is set into small, classification size increased, with normal condition away from it is larger when
There is new classification.
It is that the abnormal degree of accuracy uprises away from normal condition.On the other hand, the timing of detection can become late.
So, when classification becomes large-sized, the degree of accuracy uprises.Because loss forecasting volume can also improve therewith when the degree of accuracy is high,
So classification size and loss forecasting volume such as Fig. 8 (b) show the relation of exponential function.
In Fig. 2 step 2030, comprehensive diagnos unit 400 analyzes past data, obtains Fig. 8 relation of (b) simultaneously
It is stored in diagnostic result database 320, can also be shown in step 2040 in image display device 800 Fig. 8 (b)
Relation.
Fig. 9 is the figure of the rheological parameters' change with time for the testing result and loss forecasting volume for illustrating each diagnosis unit.
Diagnosis unit A, B, C are made up of 3 kinds of ART of different classes of size.At the moment 2200, diagnosis unit A detection, when
2210 diagnosis unit B detections are carved, at the moment 2220, diagnosis unit C detections.In addition, to the accurate of the diagnosis unit that is detected
The maximum of degree is multiplied by loss (being 10,000,000 yen in the present embodiment), counting loss amount forceasted.
So, it is varying less, required time length occurring to failure for measured value during the moment 2200-2210
State, the degree of accuracy is low and loss forecasting value is also low.As time goes by, the change of measured value becomes big, high by the degree of accuracy
Diagnosis unit detection is abnormal, and loss forecasting volume also increases.
As described above, according to the diagnostic device 200 of the present invention, can be used for according to the loss forecasting volume at each moment
Judge whether to the abnormal information handled.
Figure 10 is the figure for the modification method for illustrating the degree of accuracy.
According to the degree of failure, content, hurtful possibility changes.
For example, on the failure related to machine breakdown, tripping operation, improve the degree of accuracy and raising is carried out to loss forecasting volume and repair
Just;On the minor failure do not noticed also when inspecting periodically, reduce the degree of accuracy and loss forecasting volume is reduced
Amendment.
So, the degree of accuracy is corrected according to the disturbance degree of defect content, so as to more accurately estimate loss forecasting volume.
Figure 11 is the figure of the embodiment of picture for illustrating to show in image display device 800.
Figure 11 (a) is the implementation illustration for illustrating to be shown in picture in image display device 800 when performing diagnostic mode.
Display detects abnormal diagnosis unit and loss forecasting volume on picture.Thus, by showing damage in picture display device
Amount forceasted is lost, so as to provide the information for judging whether to processing.
Figure 11 (b) is the embodiment for illustrating to be shown in picture in image display device 800 when performing evaluation profile
Figure.It is assumed that it is the failure that can be prevented by importing diagnosis scheme than the failure that leading time detects earlier, by these failures
Loss add up and be shown as expense advantage.The expense advantage and diagnosis scheme service price calculated is shown, can interpolate that
Whether the service is bought.
Figure 12 is the figure of the embodiment of image for illustrating to show in image display device 800.
Assuming that the putting maintenance into practice when detecting failure, recommends using minimum relative to maintenance cost desired value loss forecasting volume
Diagnosis scheme.If according to the low testing result putting maintenance into practice of accuracy in detection, loss forecasting volume (risk) can reduce,
But maintenance times become more, and maintenance cost uprises.
Relative to the desired value (desired value of the maintenance cost used every year) of input maintenance cost, suitable diagnosis is exported
Scheme.So, it is relative to the desired value of input maintenance cost that also can flexibly use, and recommends the minimum diagnosis of loss forecasting volume
Scheme.
Embodiment 2
In embodiments of the invention 2, illustrate the situation for diagnosing and clustering as the application model of diagnostic techniques 500.On
The technology that cluster application describes in embodiment 1.
Figure 13 is the figure for illustrating Model Diagnosis.In Model Diagnosis, the spy of the machine of shop equipment 100 is formed using simulation
The machine mould of property.As the construction method of the pattern of simulation shop equipment 100, there are the formula of use quality conservation, heat transfer public
The statistical model such as the physical model of the physical equations such as formula and neutral net, there is Japanese Unexamined Patent Publication 2006-57595 as public technology
Number publication.
The input/output information for the machine for forming shop equipment 100 is measured respectively as signal A and signal B.In machine mould
In type, the predicted value of the signal B relative to signal A input is exported.In Model Diagnosis technology, when signal B model prediction
When value and the error of measured value exceed threshold value, exception is detected as.
Figure 14 is the figure for the effect for illustrating and being obtained with clustering technique and Model Diagnosis.
Shop equipment is connected with machine A and machine B.In diagnosing machinery B clustering technique (ART) diagnosis, by data B and
Data C detects data Long-term change trend as the input data to ART.In Model Diagnosis, relative to using data B as input,
Output data C predicted value, exception is detected as when data C predicted value and the error of measured value exceed threshold value.
In this example, at the moment 2300, there occurs the failure for being unlikely to shop equipment stopping in machine A.Because of machine A
In the influence broken down, change from machine A to machine B flow, pressure, temperature change and signal B.The moment 2300 and when
Between carving 2310, machine B regular events.At the moment 2310, because the flow that is flowed through in machine B, pressure, temperature are become
Change, so there occurs failure by machine B.
In this case, due to detecting signal B change in being diagnosed in ART, so in the timing ART at moment 2300
Diagnosis detects exception.On the other hand, because machine B is normal condition, exception is not detected in Model Diagnosis.
2310 timing, Model Diagnosis detect exception at the time of machine B breaks down.
So, ART diagnosis detect exception earlier than Model Diagnosis.In addition, the machine when detecting exception in ART
B does not break down, and when Model Diagnosis detects exception, there occurs failure by machine B.That is, detected in Model Diagnosis
The degree of accuracy that is detected during to exception is high, and in the diagnostic device 200 of the present invention, consider the degree of accuracy and by loss forecasting volume
Calculate higher.
By being obtained according to using the diagnosis unit detected as clustering technique, Model Diagnosis regularly, the degree of accuracy is different
Diagnostic result counting loss amount forceasted and show, so as to provide to judging whether that carrying out processing to the exception detected has
Use information.
Embodiment 3
Effect when illustrating for the diagnostic device 200 of the present invention to be applied to C/C shop equipments.
Figure 15 is the figure that the machine of the C/C shop equipments for the embodiment for being shown as shop equipment 1000 is formed.Combustion gas wheel
Machine 1080 is made up of compressor 1010, expanding machine 1020 and burner 1030.In gas turbine 1080, compressor 1010 is inhaled
Enter air and compress, then, burner 1030 sucks compressed air and fuel and generates burning gases, sucks expanding machine 1020
Burning gases simultaneously obtain power.The output of gas turbine 1080 is that the power that expanding machine 1020 exports and compressor reducer 1010 use
The difference of power.Heat exchanger 1060 is configured with heat recovery boiler 1050, is arranged using the high temperature from gas turbine 1080
Gas produces high-temperature steam.In steam turbine 1070, the high-temperature steam of the suction generation of heat recovery boiler 1050 simultaneously obtains power.
In condenser 1090, the exhaust of steam turbine 1070 is sucked, by the heat exchange with cooling water, steam is condensed into water.
In generator 1040, generated electricity using the output of gas turbine 1080 and steam turbine 1070.
In this shop equipment, fuel flow rate is controlled to cause delivery temperature to turn into desired value.
As the anomaly occurred in this shop equipment, can include the blade of expanding machine 1020 be used for flow through cooling
The hole (blade surface Cooling Holes) of air becomes this big phenomenon.When the anomaly occurs, cooling air is caused to become more, exhaust
Temperature reduces, the fuel flow rate increase of burner 1030.Because of the increased influence of fuel flow rate, ignition temperature rises and damages burning
Device 1030.So, the exception of expanding machine 1020 feeds through to burner 1030.
In the case that anomaly involves, as described in example 2 above, based on using detecting, timing is different with the degree of accuracy to examine
The diagnostic result that disconnected unit obtains carrys out counting loss amount forceasted and shows, thereby, it is possible to provide to judging whether to detecting
Exception carries out processing useful information.
Industrial applicability
The present invention can be extensively using the diagnostic device for shop equipment.
Claims (10)
1. a kind of shop equipment diagnostic device, possesses multiple diagnosis units of the abnormal state of diagnosis shop equipment, its feature exists
In possessing:
Comprehensive diagnos unit, it is according to the measurement signal data relevant with the state of the shop equipment and different with past state
Chang Youguan device management information data, obtain the standard of the respective detection on the abnormal state of the multiple diagnosis unit
Exactness, according to the degree of accuracy and the loss associated with abnormal state come assessment of loss amount forceasted.
2. shop equipment diagnostic device according to claim 1, it is characterised in that
Display unit is further equipped with, the display unit shows the testing result of the diagnosis unit and the loss forecasting volume.
3. shop equipment diagnostic device according to claim 1, it is characterised in that
The number for the abnormal state that the comprehensive diagnos unit passes through specified time limit divided by the shape carried out using the diagnosis unit
The abnormal detection number of state and obtain the degree of accuracy.
4. shop equipment diagnostic device according to claim 1, it is characterised in that
The device management information packet contains fault message, and the fault message includes defect content, countermeasure expense, prevented
The shop equipment of leading time, the when of breaking down required for failure stop number of days and occurred because the shop equipment stops
Opportunity loss volume.
5. shop equipment diagnostic device according to claim 1, it is characterised in that
The comprehensive diagnos unit tries to achieve average leading time, and the average leading time is to depart from base from the measurement signal data
The time when predetermined threshold for being diagnosed as abnormal state of the device management information data setting is subtracted by the detection
Unit was detected obtained from the generating state abnormal time.
6. shop equipment diagnostic device according to claim 1, it is characterised in that
The diagnosis unit is used using at least one of Model Diagnosis or cluster diagnosis diagnostic mode, the Model Diagnosis
Simulation forms the machine mould of the characteristic of the machine of shop equipment, and cluster diagnosis has used adaptive resonance theory.
7. shop equipment diagnostic device according to claim 1, it is characterised in that
The measurement signal data are categorized as multiple classifications by the multiple diagnosis unit according to similarity.
8. shop equipment diagnostic device according to claim 1, it is characterised in that
Comprehensive diagnos unit degree of accuracy according to the disturbance degree amendment of the abnormal state.
9. shop equipment diagnostic device according to claim 2, it is characterised in that
The display unit show by the desired value relative to the expense for safeguarding the abnormal state and the loss forecasting volume most
The testing result that small diagnosis unit obtains.
A kind of 10. shop equipment diagnostic method that the abnormal state of shop equipment is diagnosed using multiple methods, it is characterised in that
According to the measurement signal data relevant with the state of the shop equipment and the equipment relevant with past abnormal state
Management information data, the degree of accuracy of the respective detection on the abnormal state of the multiple method is obtained, according to the standard
Exactness and the loss associated with abnormal state carry out assessment of loss amount forceasted.
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2015124353A JP6523815B2 (en) | 2015-06-22 | 2015-06-22 | Plant diagnostic device and plant diagnostic method |
JP2015-124353 | 2015-06-22 | ||
PCT/JP2016/065373 WO2016208315A1 (en) | 2015-06-22 | 2016-05-25 | Plant diagnosis device and plant diagnosis method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107710089A true CN107710089A (en) | 2018-02-16 |
CN107710089B CN107710089B (en) | 2020-07-10 |
Family
ID=57585493
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201680036343.0A Active CN107710089B (en) | 2015-06-22 | 2016-05-25 | Plant equipment diagnosis device and plant equipment diagnosis method |
Country Status (3)
Country | Link |
---|---|
JP (1) | JP6523815B2 (en) |
CN (1) | CN107710089B (en) |
WO (1) | WO2016208315A1 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108983744A (en) * | 2017-06-05 | 2018-12-11 | 丰田自动车株式会社 | Abnormality diagnosis apparatus and abnormality diagnostic method |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6930195B2 (en) * | 2017-04-17 | 2021-09-01 | 富士通株式会社 | Model identification device, prediction device, monitoring system, model identification method and prediction method |
JP7042056B2 (en) * | 2017-10-25 | 2022-03-25 | 株式会社日立製作所 | Plant operation support device and operation support method |
WO2019239542A1 (en) | 2018-06-14 | 2019-12-19 | 三菱電機株式会社 | Abnormality sensing apparatus, abnormality sensing method, and abnormality sensing program |
JP7184636B2 (en) * | 2018-12-27 | 2022-12-06 | 三菱重工業株式会社 | Data sorting device and method, and monitoring diagnostic device |
US11361218B2 (en) | 2019-05-31 | 2022-06-14 | International Business Machines Corporation | Noise and signal management for RPU array |
JP7473343B2 (en) * | 2020-01-08 | 2024-04-23 | 株式会社日立製作所 | Operation management support device and operation management support method |
Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5566092A (en) * | 1993-12-30 | 1996-10-15 | Caterpillar Inc. | Machine fault diagnostics system and method |
JP2003295937A (en) * | 2002-03-29 | 2003-10-17 | Hitachi Ltd | Plant monitoring device and plant monitoring method |
JP2004227298A (en) * | 2003-01-23 | 2004-08-12 | Hitachi Ltd | Method for operating plant, and operation apparatus thereof |
JP2006277185A (en) * | 2005-03-29 | 2006-10-12 | Osaka Gas Co Ltd | Failure predictive diagnosis support system |
JP2007293489A (en) * | 2006-04-24 | 2007-11-08 | Mitsubishi Electric Corp | Failure diagnostic device for facility equipment and failure diagnostic method for facility equipment |
CN101194388A (en) * | 2005-06-06 | 2008-06-04 | 丰田自动车株式会社 | Error judgment device |
CN101601023A (en) * | 2007-01-31 | 2009-12-09 | 费舍-柔斯芒特系统股份有限公司 | Heat exchanger fouling detects |
CN101675195A (en) * | 2007-08-09 | 2010-03-17 | 日立建机株式会社 | Operating machine instrument diagnostic apparatus and instrument diagnostic system |
CN201714747U (en) * | 2010-06-13 | 2011-01-19 | 宝山钢铁股份有限公司 | Alarm device for hydraulic flow |
CN102036287A (en) * | 2011-01-11 | 2011-04-27 | 大唐移动通信设备有限公司 | Method and equipment for detecting abnormal state of terminal equipment |
CN102354204A (en) * | 2007-03-22 | 2012-02-15 | 日本电气株式会社 | Diagnostic device |
CN102713777A (en) * | 2010-01-22 | 2012-10-03 | 株式会社日立制作所 | Diagnostic apparatus and diagnostic method |
CN103617110A (en) * | 2013-11-11 | 2014-03-05 | 国家电网公司 | Server device condition maintenance system |
CN105122164A (en) * | 2013-04-02 | 2015-12-02 | 株式会社日立制作所 | Plant facilities management system and plant facilities management system control method |
CN105278526A (en) * | 2015-11-19 | 2016-01-27 | 东北大学 | Industrial process fault separation method on the basis of regularization framework |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011060088A (en) * | 2009-09-11 | 2011-03-24 | Toshiba Corp | Apparatus and method for preparing maintenance plan |
-
2015
- 2015-06-22 JP JP2015124353A patent/JP6523815B2/en active Active
-
2016
- 2016-05-25 WO PCT/JP2016/065373 patent/WO2016208315A1/en active Application Filing
- 2016-05-25 CN CN201680036343.0A patent/CN107710089B/en active Active
Patent Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5566092A (en) * | 1993-12-30 | 1996-10-15 | Caterpillar Inc. | Machine fault diagnostics system and method |
JP2003295937A (en) * | 2002-03-29 | 2003-10-17 | Hitachi Ltd | Plant monitoring device and plant monitoring method |
JP2004227298A (en) * | 2003-01-23 | 2004-08-12 | Hitachi Ltd | Method for operating plant, and operation apparatus thereof |
JP2006277185A (en) * | 2005-03-29 | 2006-10-12 | Osaka Gas Co Ltd | Failure predictive diagnosis support system |
CN101194388A (en) * | 2005-06-06 | 2008-06-04 | 丰田自动车株式会社 | Error judgment device |
JP2007293489A (en) * | 2006-04-24 | 2007-11-08 | Mitsubishi Electric Corp | Failure diagnostic device for facility equipment and failure diagnostic method for facility equipment |
CN101601023A (en) * | 2007-01-31 | 2009-12-09 | 费舍-柔斯芒特系统股份有限公司 | Heat exchanger fouling detects |
CN102354204A (en) * | 2007-03-22 | 2012-02-15 | 日本电气株式会社 | Diagnostic device |
CN101675195A (en) * | 2007-08-09 | 2010-03-17 | 日立建机株式会社 | Operating machine instrument diagnostic apparatus and instrument diagnostic system |
CN102713777A (en) * | 2010-01-22 | 2012-10-03 | 株式会社日立制作所 | Diagnostic apparatus and diagnostic method |
CN201714747U (en) * | 2010-06-13 | 2011-01-19 | 宝山钢铁股份有限公司 | Alarm device for hydraulic flow |
CN102036287A (en) * | 2011-01-11 | 2011-04-27 | 大唐移动通信设备有限公司 | Method and equipment for detecting abnormal state of terminal equipment |
CN105122164A (en) * | 2013-04-02 | 2015-12-02 | 株式会社日立制作所 | Plant facilities management system and plant facilities management system control method |
CN103617110A (en) * | 2013-11-11 | 2014-03-05 | 国家电网公司 | Server device condition maintenance system |
CN105278526A (en) * | 2015-11-19 | 2016-01-27 | 东北大学 | Industrial process fault separation method on the basis of regularization framework |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108983744A (en) * | 2017-06-05 | 2018-12-11 | 丰田自动车株式会社 | Abnormality diagnosis apparatus and abnormality diagnostic method |
CN108983744B (en) * | 2017-06-05 | 2021-05-14 | 丰田自动车株式会社 | Abnormality diagnosis apparatus and abnormality diagnosis method |
Also Published As
Publication number | Publication date |
---|---|
JP6523815B2 (en) | 2019-06-05 |
JP2017010232A (en) | 2017-01-12 |
WO2016208315A1 (en) | 2016-12-29 |
CN107710089B (en) | 2020-07-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107710089A (en) | Shop equipment diagnostic device and shop equipment diagnostic method | |
CN107111311B (en) | Gas turbine sensor fault detection using sparse coding methods | |
US11403160B2 (en) | Fault predicting system and fault prediction method | |
CN102999038B (en) | The diagnostic device of generating set and the diagnostic method of generating set | |
US9483049B2 (en) | Anomaly detection and diagnosis/prognosis method, anomaly detection and diagnosis/prognosis system, and anomaly detection and diagnosis/prognosis program | |
CN106404403B (en) | Method and system for analysis of a turbomachine | |
JP6856443B2 (en) | Equipment abnormality diagnosis system | |
JP5561835B2 (en) | System and method for article monitoring | |
CN102182671B (en) | State analysis monitoring method of gas compressor | |
CN102713777B (en) | Diagnostic apparatus and diagnostic method | |
US20190130294A1 (en) | System fault isolation and ambiguity resolution | |
GB2476246A (en) | Diagnosing an operation mode of a machine | |
JP2012118981A (en) | Turbine performance diagnostic system and methods | |
KR102073537B1 (en) | Risk Assessment Apparatus, Risk Assessment System, Risk Assessment Method and Risk Assessment Program | |
McKee et al. | A review of machinery diagnostics and prognostics implemented on a centrifugal pump | |
JP6915693B2 (en) | System analysis method, system analyzer, and program | |
JP6830414B2 (en) | Diagnostic device and diagnostic method | |
KR20200009081A (en) | Risk Assessment Apparatus, Risk Assessment System, Risk Assessment Method, Risk Assessment Program and Data Structure | |
JP6685124B2 (en) | Diagnostic device and diagnostic method | |
Beduschi et al. | Optimizing rotating equipment maintenance through machine learning algorithm | |
King et al. | Probabilistic approach to the condition monitoring of aerospace engines | |
JP6482742B1 (en) | Risk assessment device, risk assessment system, risk assessment method, and risk assessment program | |
KR102212022B1 (en) | Method of automatically determining condition of hydro turbine in hydroelectric power plant and system for the same | |
JP2018055391A (en) | Security diagnostic device and security diagnostic method | |
CN115668085A (en) | Plant monitoring device, plant monitoring method, and program |
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 |