WO2016208315A1

WO2016208315A1 - Plant diagnosis device and plant diagnosis method

Info

Publication number: WO2016208315A1
Application number: PCT/JP2016/065373
Authority: WO
Inventors: 孝朗関合; 林　喜治; 達矢前田; 和貴定江; 正博村上; 深井　雅之
Original assignee: 株式会社日立製作所
Priority date: 2015-06-22
Filing date: 2016-05-25
Publication date: 2016-12-29
Also published as: JP6523815B2; CN107710089B; CN107710089A; JP2017010232A

Abstract

When an abnormality is detected, a decision about whether or not the abnormality should be dealt with is made based onon the basis of experiences in operating a plant. However, it is preferable to make the decision on the basis ofbased on a risk (expected loss amount) that would occur when if the abnormality iswere to be left as is. In order to achieve this, the present invention is a plant diagnosis device equipped with multiple diagnosis means for diagnosing an abnormality state of the plant, and characterized by being equipped with a comprehensive diagnosis means that, upon determining degrees of accuracy in detecting the abnormal states with respect to each of the multiple diagnosis means on the basis of measurement signal data pertaining to plant states and facility management information data pertaining to past abnormal states, evaluates expected loss amounts on the basis of the degrees of accuracy and loss amounts associated with the abnormal states.

Description

Plant diagnostic device and plant diagnostic method

The present invention relates to a plant diagnostic apparatus and a plant diagnostic method for diagnosing abnormal plant conditions.

The plant diagnostic device detects the occurrence of an abnormality or accident based on the measurement data from the plant when an abnormal transient or accident occurs in the plant.

Patent Document 1 discloses a diagnostic apparatus using adaptive resonance theory (ART), which is one of clustering technologies. Here, ART is a theory that classifies multidimensional data into categories according to their similarity.

In the above technology, first, normal measurement data is classified into a plurality of categories (normal categories) using ART. Next, the current measurement data is input to the ART and classified into categories. When this measurement data cannot be classified into normal categories, a new category (new category) is generated. The occurrence of a new category means that the state of the plant has changed. Therefore, the occurrence of an abnormality is determined based on the occurrence of a new category, and an abnormality is diagnosed when the occurrence rate of the new category exceeds a threshold value.

JP 2005-165375 A

In the clustering technology, there is a parameter that determines the size of the cluster (the size of the category in ART). This parameter is called a resolution parameter. Generally, when classifying certain data into clusters, if the resolution is set coarsely, the number of clusters decreases, and if the resolution is set finely, the number of clusters increases.

際 When using clustering for abnormality diagnosis, the range of change in the data trend in which new categories occur differs depending on whether the resolution is coarse or fine. When a new category is generated when the resolution is rough, the data tendency is greatly changed from that in the normal state, so the probability that the devices are different is high. On the other hand, when the resolution is fine, there is a possibility that a minute tendency change such as measurement noise is detected, so the probability of abnormality is low. Thus, when the setting values of the parameters that determine the size of the cluster are different, the probability that an abnormality has occurred at the time of detecting an abnormality differs.

Generally, since the abnormality detection performance differs when the diagnostic method is different, the probability that an abnormality has occurred at the time of abnormality detection is different.

Also, it is effective to stop and maintain and repair the plant when an abnormality is detected to avoid equipment failure. However, costs for maintenance and repair, and opportunity loss during the period when the plant is stopped occur. Therefore, if there is a minor failure, the operation may be continued until a regular inspection. On the other hand, as a result of neglecting the abnormality, the device may break down or be damaged, and there is a possibility that the loss will be larger than when maintenance or repair is performed.

<Currently, when an abnormality is detected, whether or not to deal with the abnormality is determined based on the experience of plant operation, but it is desirable to determine based on the risk (loss expected amount) when the abnormality is left unattended.

In order to solve the above-mentioned problems, the present invention provides a plant diagnostic apparatus comprising a plurality of diagnostic means for diagnosing a plant state abnormality, based on measurement signal data relating to the plant state and facility management information data relating to a past state abnormality. And a comprehensive diagnosis unit that obtains an accuracy of detection of the state abnormality of each of the plurality of diagnosis units and evaluates an estimated loss amount based on the accuracy and a loss amount associated with the state abnormality.

求め Estimates the estimated amount of loss when an abnormality is detected, and can provide useful information for determining whether to handle the detected abnormality.

It is a block diagram explaining the diagnostic apparatus which is 1st Example of this invention. It is a flowchart figure explaining operation | movement of the comprehensive diagnostic means in evaluation mode and diagnostic mode of a diagnostic apparatus. It is a figure explaining the timing which operates evaluation mode and diagnostic mode. It is a figure explaining the aspect of the data preserve | saved at a measurement signal database and an equipment management information database. It is a figure explaining the aspect of the data preserve | saved at a diagnostic result database. It is explanatory drawing of an adaptive resonance theory. It is a figure which shows the example of a result which classified the measurement signal into the category. It is a figure explaining the relationship between the size of a category, detection timing, accuracy, and an estimated loss amount. It is a figure explaining the time-dependent change of the detection result of each diagnostic means, and an estimated loss amount. It is a figure explaining the correction method of accuracy. It is a figure explaining the Example of the screen displayed on a screen display apparatus. It is a figure explaining the Example of the screen displayed on a screen display apparatus. It is a figure explaining model diagnosis. It is a figure explaining the effect by using clustering diagnosis and model diagnosis together. It is a figure explaining the Example at the time of applying the diagnostic apparatus of this invention to a thermal power plant.

A diagnostic apparatus suitable for implementing the present invention will be described below with reference to the drawings. It should be noted that the following is merely an example of implementation and is not intended to limit the invention itself to the following specific contents.

FIG. 1 is a block diagram for explaining a diagnostic apparatus according to a first embodiment of the present invention. The diagnosis device 200 is connected to the plant 100, the screen display device 800, and the external input device 900, and monitors and diagnoses the plant 100. In addition, the diagnostic device 200 is configured by connecting a communication unit that performs communication between devices or devices, a computer, a computer server (CPU: Central Processing Unit), a memory, various database DBs, and the like by wired or wireless connection. The external input device 900 includes a keyboard switch, a pointing device such as a mouse, a touch panel, a voice instruction device, and the like, and the screen display device 800 includes a display.

The diagnostic device 200 includes a comprehensive diagnostic unit 400 and a diagnostic unit 500 as arithmetic units. A plurality of diagnosis means 500 are provided, and the number thereof can be arbitrarily set. The diagnostic apparatus 200 includes a measurement signal database 300, an equipment management information database 310, and a diagnostic result database 320 as databases. In FIG. 1, the database is abbreviated as DB.

The computerized information is stored in the measurement signal database 300, the facility management information database 310, and the diagnosis result database 320, and the information is stored in a form generally called an electronic file (electronic data).

Moreover, the diagnostic apparatus 200 includes an external input interface 210 and an external output interface 220 as interfaces with the outside.

And the measurement signal 1 which measured various state quantities which are the operating states of the plant 100 via the external input interface 210, and the external input signal 2 created by operation of the keyboard 910 and the mouse 920 provided in the external input device 900 Is taken into the diagnostic apparatus 200. Further, the comprehensive diagnosis result signal 12 is output to the screen display device 800 via the external output interface 220.

In the diagnostic apparatus 200 shown in FIG. 1, the measurement signal 1 obtained by measuring various state quantities of the plant 100 is taken in via the external input interface 210. The measurement signal 3 captured by the diagnostic device 200 is stored in the measurement signal database 300. Also, facility management information such as failure information and maintenance information generated in the plant 100 is taken into the diagnostic apparatus 200 by an external input signal 2 generated by operating the keyboard 910 and the mouse 920. The facility management information signal 4 captured by the diagnostic device 200 is stored in the facility management information database 310.

Diagnostic device 200 has two processing modes, an evaluation mode and a diagnostic mode. The flowchart of the evaluation mode and the diagnosis mode and the operations of the comprehensive diagnosis unit 400 and the diagnosis unit 500 will be described later with reference to FIGS.

In the diagnostic apparatus 200 of the present embodiment, the comprehensive diagnostic unit 400, the diagnostic unit 500, the measurement signal database 300, the equipment management information database 310, and the diagnostic result database 320 are provided in the diagnostic apparatus 200. May be arranged outside the diagnostic apparatus 200, and only data may be communicated between the apparatuses.

Further, all the information stored in the database installed in the diagnostic device 200 can be displayed on the screen display device 100, and these information are the external input signal 1 generated by operating the external input device 900. It can be corrected.

In this embodiment, the external input device 900 is composed of a keyboard and a mouse, but any device for inputting data such as a microphone or a touch panel for voice input may be used.

Further, it goes without saying that as an embodiment of the present invention, the present invention can also be implemented as an information providing service for providing information obtained by operating the diagnostic method and diagnostic apparatus 200.

FIG. 2 is a flowchart for explaining the operation of the comprehensive diagnosis means 400 in the evaluation mode and the diagnosis mode of the diagnosis apparatus 200.

FIG. 2 (a) is a flowchart of the evaluation mode.

First, in step 2000, the comprehensive diagnosis means 400 extracts the measurement signal 5 during a predetermined period stored in the measurement signal database 300.

In step 2010, the comprehensive diagnosis unit 400 transmits the measurement signal 9 to the diagnosis unit 500. The diagnosis unit 500 processes the measurement signal 9 to diagnose the state of the plant 100 and transmits the diagnosis result 10 to the comprehensive diagnosis unit 400. The comprehensive diagnosis unit 400 collects the received diagnosis results 10 and transmits the diagnosis result database information 8 to the diagnosis result database 320 for storage.

In step 2020, the comprehensive diagnosis unit 400 extracts the facility management information signal 6 stored in the facility management information database 310.

In step 2030, the detection result of each diagnostic means in the diagnosis result database information 7 stored in the diagnosis result database 320 is compared with the facility management information signal 6 extracted in step 2020, and the accuracy and average lead time are calculated. Here, the accuracy is obtained by dividing the number of failures by the number of detections. The average lead time is a time obtained by subtracting the time detected by the corresponding diagnostic means from the time detected by the threshold determination, and is a time indicating how early the time is detected. The accuracy and average lead time of each diagnostic means obtained in step 2030 are stored in the diagnostic result database 320.

In step 2040, the comprehensive diagnosis means 400 extracts the diagnosis result database information 7 stored in the diagnosis result database 320 and transmits it to the external output interface 220 as the comprehensive diagnosis result signal 11. The comprehensive diagnosis result signal 12 is transmitted to the image display device 800 and displayed on the screen display device 800.

FIG. 2B is a flowchart for explaining the operation in the diagnosis mode.

In step 2100, the comprehensive diagnosis means 400 extracts the operation data 5 for the period to be diagnosed that is stored in the measurement signal database.

In step 2110, the comprehensive diagnosis unit 400 transmits the measurement signal 9 to the diagnosis unit 500. The diagnosis unit 500 processes the measurement signal 9 to diagnose the state of the plant 100 and transmits the diagnosis result 10 to the comprehensive diagnosis unit 400. The comprehensive diagnosis unit 400 collects the received diagnosis results 10 and transmits the diagnosis result database information 8 to the diagnosis result database 320 for storage.

In step 2120, the presence / absence of abnormality detection is evaluated. If there is a diagnostic means that has detected the abnormality, the process proceeds to step 2130, and if not, the process proceeds to step 2160.

In step 2130, the comprehensive diagnosis unit 400 extracts the diagnosis result database information 7 stored in the diagnosis result database 320, and grasps the accuracy information regarding the diagnosis unit that detected the abnormality in step 2120.

In step 2140, the comprehensive diagnosis unit 400 extracts the facility management information 6 stored in the facility management information database 310, and grasps the loss due to the failure.

In step 2150, the comprehensive diagnosis unit 400 calculates the estimated loss based on the accuracy extracted in step 2130 and the loss extracted in step 2140. Needless to say, the expected loss amount may be obtained in a plurality of ways, such as by multiplying the accuracy and the loss amount, or by using a predetermined parameter.

In Step 2160, the detection result of each diagnostic means and the estimated loss calculated in Step 2150 when there is a diagnostic means that has detected an abnormality are displayed on the screen display device 800.

Thus, in the diagnostic apparatus 200 of the present invention, when the abnormality is detected by the diagnostic unit 500, it is possible to provide useful information for determining whether or not to deal with the detected abnormality by displaying the estimated loss amount.

FIG. 3 is a diagram for explaining the timing for operating the evaluation mode and the diagnostic mode.

3A, after accumulating operation data for a certain period, the evaluation mode is operated once and the diagnosis mode is operated at a certain period.

In the method shown in FIG. 3B, the evaluation mode is operated at regular intervals, the accuracy and the average lead time data stored in the diagnosis result database 320 are updated, and then the diagnosis mode is operated.

In the method shown in FIG. 3 (c), the evaluation mode is activated when an instruction is given from the user. The evaluation mode is executed at an arbitrary timing, the accuracy and the average lead time are updated, and the diagnosis mode is operated.

In addition to the timing described in the present embodiment, the timing for operating the evaluation mode and the diagnostic mode can be arbitrarily set.

FIG. 4 is a diagram for explaining a mode of data stored in the measurement signal database 300 and the facility management information database 310.

As shown in FIG. 4A, in the measurement signal database 300, the value of the measurement signal 1 (data items A, B, and C are shown in the figure) that is operation data measured for the plant 100 is sampled. Stored for each period (time on the vertical axis).

By using

scroll boxes

302 and 303 that can move vertically and horizontally on the display screen 301, a wide range of data can be scroll-displayed.

As shown in FIG. 4B, the facility management information database 310 stores failure information such as failure contents, countermeasure costs, lead time required for failure avoidance, days of stoppage due to failure, and opportunity loss caused by plant shutdown. Is done.

Further, as shown in FIG. 4C, maintenance information such as maintenance contents, cost required for maintenance, the number of days required for maintenance, an opportunity loss due to maintenance, and the like are stored in the facility management information database 310.

FIG. 5 is a diagram for explaining a mode of data stored in the diagnosis result database 320.

As shown in FIG. 5A, in the diagnosis result database 320, detection results of the respective diagnosis means (diagnosis means A, B, and C are described in the figure) are stored for each sampling period (time on the vertical axis). Is done.

By using scroll boxes 312 and 313 that can be moved vertically and horizontally on the display screen 311, a wide range of data can be scroll-displayed.

In the diagnosis result database 320, the detection result of each diagnosis means is stored. For example, the diagnosis result is replaced with digital information, such as 1 at the time of abnormality determination and 0 at the time of normal determination.

As shown in FIG. 5 (b), in the diagnosis result database, the accuracy and average lead time calculated in the evaluation mode are stored for each diagnosis means.

FIG. 6 describes a case where an adaptive resonance theory (ART) is applied as an example of the diagnostic means 500. It should be noted that other clustering methods such as vector quantization and support vector machine can be used.

As shown in FIG. 6A, the data classification function includes a data preprocessing device 610 and an ART module 620. The data preprocessing device 610 converts the operation data into input data for the ART module 620.

Hereinafter, those procedures performed by the data preprocessing device 610 and the ART module 620 will be described.

First, the data pre-processing device 610 normalizes the data for each measurement item. Data including the data Nxi (n) obtained by normalizing the measurement signal and the complement CNxi (n) (= 1−Nxi (n)) of the normalized data is defined as input data Ii (n). This input data Ii (n) is input to the ART module 620.

In the ART module 620, the measurement signal 10 or the operation signal 11 which is input data is classified into a plurality of categories.

The ART module 620 includes an F0 layer 621, an F1 layer 622, an F2 layer 623, a memory 624, and a selection subsystem 625, which are coupled to each other. The F1 layer 622 and the F2 layer 623 are connected via a weighting factor. The weighting coefficient represents a prototype (prototype) of a category into which input data is classified. Here, the prototype represents a representative value of the category.

Next, the algorithm of the ART module 620 will be described.

The outline of the algorithm when input data is input to the ART module 620 is as shown in Process 1 to Process 5 below.

Process 1: The input vector is normalized by the F0 layer 621, and noise is removed.

Process 2: A suitable category candidate is selected by comparing the input data input to the F1 layer 622 with a weighting factor.

Process 3: The validity of the category selected by the selection subsystem 625 is evaluated by the ratio with the parameter ρ. If it is determined to be valid, the input data is classified into the category, and the process proceeds to process 4. On the other hand, if it is not determined to be valid, the category is reset, and an appropriate category candidate is selected from the other categories (repeat process 2). Increasing the value of parameter ρ makes the category classification finer. That is, the category size is reduced. Conversely, if the value of ρ is reduced, the classification becomes coarse. Category size increases. This parameter ρ is referred to as a vigilance parameter.

Process 4: When all the existing categories are reset in Process 2, it is determined that the input data belongs to the new category, and a new weighting factor representing the prototype of the new category is generated.

Process 5: When input data is classified into category J, weight coefficient WJ (new) corresponding to category J uses past weight coefficient WJ (old) and input data p (or data derived from input data). Is updated by equation (1).

(Equation 1)
WJ (new) = Kw · p + (1-Kw) · WJ (old)

Here, Kw is a learning rate parameter (0 <Kw <1), and is a value that determines the degree to which the input vector is reflected in the new weighting factor.

It should be noted that

Equations

1 and 2 to 12 described below are incorporated in the ART module 620.

The characteristic of the data classification algorithm of the ART module 620 is the above processing 4.

In process 4, when input data different from the learned pattern is input, a new pattern can be recorded without changing the recorded pattern. Therefore, it is possible to record a new pattern while recording a pattern learned in the past.

Thus, when the operation data given in advance is given as input data, the ART module 620 learns the given pattern. Therefore, when new input data is input to the learned ART module 620, it is possible to determine which pattern in the past is close by the above algorithm. If the pattern has never been experienced before, it is classified into a new category.

FIG. 6B is a block diagram showing the configuration of the F0 layer 621. In the F0 layer 621, the input data I _i is normalized again at each time, and a normalized input vector u _i ⁰ to be input to the F1 layer 621 and the selection subsystem 625 is created.

First, W _i ⁰ is calculated from the input data I _i according to Equation 2. Here, a is a constant.

Next, X _i ⁰ obtained by normalizing W _i ⁰ is calculated using Equation 3. Here, || W ⁰ || represents the norm of W ⁰ .

Then, using equation 4, to calculate the V _i ⁰ obtained by removing noise from the X _i ^0. However, θ is a constant for removing noise. Since the minute value becomes 0 by the calculation of Equation 4, noise of the input data is removed.

Finally, a normalized input vector u _i ⁰ is obtained using Equation 5. u _i ⁰ is input to the F1 layer.

FIG. 6C is a block diagram showing the configuration of the F1 layer 622. As shown in FIG. In the F1 layer 622, u _i ⁰ obtained by Expression 5 is held as a short-term memory, and P _i to be input to the F2 layer 722 is calculated. Formulas for the F2 layer are collectively shown in Formulas 6 to 12. However, a and b are constants, f (•) is a function expressed by Equation 4, and T _j is a fitness calculated by the F2 layer 623.

However,

FIG. 7 is a diagram showing an example of the result of classifying measurement signals into categories.

FIG. 7A is a diagram illustrating an example of a classification result obtained by classifying the measurement signal 1 of the plant 100 into categories.

FIG. 7 (a) shows, as an example, two items of the measurement signal, which are represented by a two-dimensional graph. In addition, the vertical axis and the horizontal axis indicate the measurement signals of the respective items normalized.

The measurement signal is divided into a plurality of categories 630 (circles shown in FIG. 4C) by the ART module 620 in FIG. One circle corresponds to one category.

In this embodiment, measurement signals are classified into four categories. Category number 1 is a group in which item A has a large value and item B has a small value, category number 2 is a group in which both items A and B have small values, category number 3 has a small value in item A, and item B A group with a large value of, category number 4 is a group with a large value for both items A and B.

FIG. 7B is a diagram for explaining the result of classifying the measurement signal 1 acquired from the plant 100 into categories. The horizontal axis represents time, and the vertical axis represents measurement signals and category numbers.

As shown in FIG. 7 (b), the data of the normal period before the start of diagnosis was classified into categories 1 to 3. The first half of data after the start of monitoring is classified into category 2, which is the same category as the model data. In this case, since the data trends are the same, it is determined that the state has not changed. On the other hand, the latter half of data after the start of monitoring is classified into category 4, and is classified into a category different from the model data. Since the data trends are different, it is determined that the state of the plant has changed.

Thus, diagnostic technology using clustering technology has the feature of detecting changes in data trends.

FIG. 8 is a diagram for explaining the relationship between the category size, detection timing, accuracy, and expected loss.

As shown in FIG. 8A, if the parameter ρ for determining the resolution is set large and the category size is reduced, even a small change is detected. Early detection is possible. On the other hand, accuracy is lowered because minute changes such as measurement noise are detected.

On the other hand, if the parameter ρ is set small and the category size is increased, a new category is generated when the deviation from the normal state is large.

大きく It is far from the normal state, and the probability of abnormality is high. On the other hand, the detection timing is delayed.

Thus, the accuracy increases as the category size increases. If the accuracy is high, the estimated loss amount is also high, so the category size and the estimated loss amount have an exponential relationship as shown in FIG.

In step 2030 of FIG. 2, past data is analyzed by the comprehensive diagnosis means 400 to obtain the relationship of FIG. 8B and stored in the diagnosis result database 320. In step 2040, the relationship of FIG. You may make it display on the apparatus 800. FIG.

FIG. 9 is a diagram for explaining the change over time in the detection results of each diagnostic means and the estimated loss amount.

Diagnostic means A, B, C are composed of three types of ART with different category sizes. Diagnosis means A detects at time 2200, diagnosis means B detects at time 2210, and diagnosis means C detects at time 2220. In addition, the estimated loss amount is calculated by multiplying the maximum value of the accuracy of the detected diagnostic means by the amount of damage (10 million yen in this embodiment).

Thus, during the time 2200-2210, the change in the measured value is small, the time to reach the failure is long, the accuracy is low, and the expected loss is also low. The change in the measured value increases with the passage of time, an abnormality is detected by a highly accurate diagnostic means, and the expected loss is also increased.

As described above, the diagnosis device 200 of the present invention can acquire information for determining whether to deal with an abnormality based on the estimated loss at each time.

FIG. 10 is a diagram for explaining an accuracy correction method.

The possibility of damage varies depending on the degree and content of the failure.

For example, for failures that lead to equipment damage or trips, increase the accuracy to correct the expected loss, and for minor failures that were not noticed until the regular inspection, reduce the accuracy and reduce the expected loss to a lower level.

In this way, it is possible to estimate the estimated loss more accurately by correcting the accuracy according to the degree of influence of the failure content.

FIG. 11 is a diagram for explaining an example of a screen displayed on the screen display device 800.

FIG. 11A is a diagram illustrating an example of a screen displayed on the screen display device 800 when the diagnosis mode is executed. The diagnostic means that detected the abnormality and the estimated loss amount are displayed on the screen. In this way, by displaying the expected loss amount on the screen display device, it is possible to provide information for determining whether or not to deal with it.

FIG. 11B illustrates an example of a screen displayed on the screen display device 800 when the evaluation mode is executed. Assuming that a failure detected earlier than the lead time is a failure that can be prevented by introducing a diagnostic plan, the loss amount of these failures is added and displayed as a cost merit. It is possible to display the calculated cost merit and the service price of the diagnostic plan and determine whether or not to purchase this service.

FIG. 12 is a diagram for explaining an example of a screen displayed on the screen display device 800.

Assuming that maintenance is performed when detected, a diagnosis plan that minimizes the expected loss for the maintenance cost target value is proposed. If maintenance is performed with a detection result of a diagnosis with low wrinkle detection accuracy, the expected loss (risk) can be reduced, but the number of wrinkle maintenance increases and the maintenance cost increases.

Suppose that the target value of maintenance cost (target value of annual maintenance cost) is input, and a suitable diagnosis plan is output. In this way, the system can be utilized as a system for proposing a diagnosis plan that minimizes the expected loss for the input of the maintenance cost target value.

In the second embodiment of the present invention, a case where model diagnosis and clustering are used as the diagnosis technique 500 will be described. For the clustering, the technique described in the first embodiment is used. *

FIG. 13 is a diagram for explaining model diagnosis. In the model diagnosis, a device model that simulates the characteristics of the devices constituting the plant 100 is used. As a method for constructing a model that simulates the plant 100, there are a physical model using a physical equation such as a mass conservation equation, a heat transfer equation, and a statistical model such as a neural network, and Japanese Patent Application Laid-Open No. 2006-57595 is known. .

Measure input / output information of devices constituting the plant 100 as signal A and signal B, respectively. In the device model, a predicted value of the signal B with respect to the input of the signal A is output. In the model diagnosis technique, an abnormality is detected when an error between a model predicted value and an actual measurement value of the signal B exceeds a threshold value.

FIG. 14 is a diagram for explaining the effect of using clustering and model diagnosis together.

In the plant, equipment A and equipment B are connected. In clustering (ART) diagnosis for diagnosing device B, data B and data C are used as input data to the ART, and a change in the data trend is detected. In model diagnosis, data B is input, a predicted value of data C is output, and an abnormality is detected when an error between the predicted value of data C and an actual measured value exceeds a threshold value.

In this case, at time 2300, a trouble that equipment A did not stop the plant occurred. The flow rate, pressure, and temperature flowing from the device A to the device B change due to the trouble that has occurred in the device A, and the signal B changes. Between time 2300 and time 2310, device B is operating normally. Trouble occurred in device B due to changes in the flow rate, pressure, and temperature of the fluid flowing to device B at time 2310.

In this case, since the change of the signal B is detected by the ART diagnosis, the ART diagnosis detects an abnormality at the timing of time 2300. On the other hand, since the device B is in a normal state, no abnormality is detected in the model diagnosis.

The model diagnosis is detected at the timing of time 2310 when trouble occurs in device B.

In this way, ART diagnosis detects an abnormality earlier than model diagnosis. Further, no trouble has occurred in the device B when detected by ART, and a trouble has occurred when detected by model diagnosis. That is, the probability of abnormality is higher when detected by model diagnosis, and the diagnosis device 200 of the present invention calculates the estimated loss amount in consideration of this accuracy.

Information useful for determining whether or not to handle detected abnormalities by calculating and displaying the expected loss based on the diagnosis results using diagnostic means with different detection timing and accuracy, such as clustering and model diagnosis Can provide.

The effect when the diagnostic apparatus 200 of the present invention is applied to a C / C plant will be described.

FIG. 15 is a diagram showing a device configuration of a C / C plant which is an embodiment of the plant 1000. As shown in FIG. The gas turbine 1080 includes a compressor 1010, an expander 1020, and a combustor 1030. In the gas turbine 1080, the compressor 1010 takes in air and compresses it, then the combustor 1030 takes in compressed air and fuel to generate combustion gas, and the expander 1020 takes in combustion gas to obtain power. The output of the gas turbine 1080 is the difference between the power output from the expander 1020 and the power used by the compressor 1010. The exhaust heat recovery boiler 1050 is provided with a heat exchanger 1060 and generates high temperature steam using the high temperature exhaust gas from the gas turbine 1080. In the steam turbine 1070, the high-temperature steam generated by the exhaust heat recovery boiler 1050 is taken in to obtain power. In the condenser 1090, the exhaust of the steam turbine 1070 is taken in and heat-exchanged with the cooling water to condense the steam into water. The generator 1040 generates power using the outputs of the gas turbine 1080 and the steam turbine 1070.

In this plant, the fuel flow rate is controlled so that the exhaust gas temperature becomes the target value.

An abnormal event that occurs in this plant is that the holes (blade surface cooling holes) for supplying cooling air for the blades in the expander 1020 become very large. When this abnormality occurs, the amount of cooling air increases, the exhaust gas temperature decreases, and the fuel flow rate of the combustor 1030 increases. The combustion temperature rises due to the increase in the fuel flow rate, and the combustor 1030 is damaged.異常 Thus, the abnormality of the expander 1020 is spread to the combustor 1030.

If an abnormal event spreads, as described in Example 2, is it possible to deal with the detected abnormality by calculating and displaying the estimated loss based on the diagnosis result using the diagnostic means with different detection timing and accuracy? It can provide information useful for determining whether or not.

The present invention is widely applicable as a plant diagnostic apparatus.

DESCRIPTION OF SYMBOLS 1 Measurement signal 2 External input signal 3 Measurement signal 4 Equipment management information signal 5 Measurement signal 6 Equipment management information signal 7 Diagnosis result database information 8 Diagnosis result database information 9 Measurement signal 10 Diagnosis result 11 Comprehensive diagnosis result signal 12 Comprehensive diagnosis result signal 100 Plant 200 Diagnostic device 210 Data input interface 220 Data output interface 300 Measurement signal database 310 Facility management information database 320 Diagnosis result database 400 Overall diagnosis means 500 Diagnosis means 800 Surface display device 900 External input device 910 Keyboard 920 Mouse

Claims

In a plant diagnostic apparatus comprising a plurality of diagnostic means for diagnosing abnormal plant conditions,
Based on the measurement signal data relating to the state of the plant and the facility management information data relating to the past state abnormality, the accuracy relating to the detection of the state abnormality of each of the plurality of diagnostic means is obtained, and the loss associated with the accuracy and the state abnormality is obtained. A plant diagnostic apparatus comprising comprehensive diagnostic means for evaluating an estimated loss amount based on the total diagnostic means.
The plant diagnostic apparatus according to claim 1 is:
A plant diagnosis apparatus, further comprising display means for displaying the detection result of the diagnosis means and the estimated loss amount.
The plant diagnostic apparatus according to claim 1,
The plant diagnosis apparatus characterized in that the comprehensive diagnosis unit obtains the accuracy by dividing the number of state abnormalities in a predetermined period by the number of state abnormality detections by the diagnostic unit.
The plant diagnostic apparatus according to claim 1,
The facility management information data includes failure information composed of failure contents, countermeasure costs, lead time necessary for preventing failure occurrence, the number of days of plant shutdown in the event of failure, and the opportunity loss amount caused by the plant shutdown A plant diagnostic apparatus characterized by that.
The plant diagnostic apparatus according to claim 1,
The comprehensive diagnosis means detects that a state abnormality has occurred by the diagnosis means from a time when the measurement signal data deviates from a predetermined threshold value that is diagnosed as a state abnormality set based on the facility management information data. A plant diagnostic apparatus characterized in that an average lead time obtained by subtracting the measured time is obtained.
The plant diagnostic apparatus according to claim 1,
The plant diagnosis apparatus characterized in that the diagnosis means uses at least one diagnosis of a model diagnosis using an equipment model that simulates characteristics of equipment constituting the plant or a clustering diagnosis using an adaptive resonance theory.
The plant diagnostic apparatus according to claim 1,
The plurality of diagnosis means classifies the measurement signal data into a plurality of categories according to the degree of similarity.
The plant diagnostic apparatus according to claim 1,
The plant diagnosis apparatus, wherein the comprehensive diagnosis unit corrects the accuracy according to an influence degree of the state abnormality.
The plant diagnostic apparatus according to claim 2,
The plant diagnosis apparatus, wherein the display means displays a detection result by a diagnosis means that minimizes the expected loss for a target value of a cost for maintaining the state abnormality.
In a plant diagnosis method for diagnosing abnormal plant conditions by multiple methods,
Based on the measurement signal data relating to the state of the plant and the facility management information data relating to the past state abnormality, the accuracy relating to the detection of the state abnormality of each of the plurality of methods is obtained, and based on the accuracy and the loss amount associated with the state abnormality. A plant diagnosis method characterized by evaluating an estimated loss amount.