JP5164954B2 - Device diagnostic method and device diagnostic device - Google Patents

Description

  The present invention relates to a device diagnosis method and a device diagnosis device, and more particularly to a device diagnosis method and a device diagnosis device suitable for application to a nuclear power plant.

  Devices such as sensors, rotating machines and valves in the nuclear power plant are subjected to maintenance work every periodic inspection of the nuclear power plant. Equipment maintenance work is mainly performed in periodic inspections, and the amount of equipment maintenance work such as calibration, overhaul and replacement is enormous.

  For the purpose of operating nuclear power plants safely and efficiently, the introduction of state monitoring maintenance that monitors the state of equipment and performs maintenance work when necessary, instead of the time maintenance in which maintenance work is performed at regular intervals. Is being considered.

  Equipment status monitoring is performed by sensors such as a thermometer and a vibration meter attached to the equipment. For this reason, if a drift occurs in the sensor (detector) and the measured value deviates from the correct value, the state of the device cannot be accurately diagnosed. When any abnormality is recognized in the process value of the device measured by the sensor, it is necessary to distinguish between the device abnormality and the sensor drift in order to determine whether the device should be maintained or the sensor should be maintained. Further, when some abnormality is detected, even if the sensor drift is the cause, if it cannot be distinguished, there is a possibility that the nuclear power plant will be stopped. Therefore, the distinction between device abnormality and sensor drift is an important issue.

  An example of a device diagnosis method is described in JP-A-6-167363. In this device diagnosis method, production rules and membership functions are used. The relationship between the magnitude of the device state quantity and the degree of occurrence of multiple abnormal phenomena occurring in the equipment is expressed by production rules and membership functions, and the measured state quantities are referred to the production rules and membership functions for fuzzy calculations. Then, the degree of conformity is obtained by quantifying the degree of occurrence of a plurality of abnormal phenomena. In this way, the type of abnormality that has occurred in the device is specified.

  Japanese Patent Laying-Open No. 2003-207373 describes an example of a sensor calibration support method. This calibration support method uses detection signals output from a plurality of sensors that are provided in a detection target and have a correlation with each other. Using an estimation model for estimating the true value prepared according to the detection target, the true value is estimated based on the actual measurement value of the detection signal, and based on the actual measurement value and the estimated true value, Estimate the drift amount over the full span. A sensor calibration amount is obtained based on the estimated drift amount.

  Japanese Patent Laying-Open No. 2005-43121 describes an instrument calibration support system. This system predicts the calibration cycle of the sensor using the calibration data of the sensor, and obtains the calibration cycle of the sensor based on the prediction result and the result of online drift detection.

JP-A-6-167363 JP 2003-207373 A JP-A-2005-43121

  When diagnosing the presence / absence of device abnormality, as described above, it is necessary to distinguish between device abnormality and sensor drift. Even if the device diagnosis method described in JP-A-6-167363 and JP-A-2003-207373 are combined with a sensor calibration support method, it is not possible to distinguish between device abnormality and sensor drift. If the measurement value shifts due to sensor drift, a misdiagnosis may occur in device diagnosis. Conversely, there is a possibility that a change in the process value due to device abnormality is erroneously diagnosed as sensor drift. As a result, there is a possibility of diagnosing an abnormality in either device diagnosis or sensor diagnosis, but it is difficult to distinguish which is a wrong diagnosis.

  In the device diagnosis method described in Japanese Patent Application Laid-Open No. 6-167363, the type of abnormality can be distinguished. Therefore, by measuring in advance the data when each abnormality of the device abnormality and the sensor drift occurs, both Can be distinguished. However, in a nuclear power plant, it is often difficult to measure in advance data when an abnormality occurs.

  An object of the present invention is to provide a device diagnosis method and a device diagnosis apparatus that can improve the accuracy of device abnormality determination.

  The feature of the present invention that achieves the above-described object is calculated by calculating the drift amount of measurement data measured by a sensor provided in the plant and obtaining the drift distribution of the sensor based on a plurality of calibration data of the sensor. Output the measurement data used to calculate the drift amount when it is determined that the calculated drift amount is within the standard deviation of the drift distribution. The measurement data of the sensor is corrected using the drift amount, and the equipment provided in the plant is diagnosed based on the corrected measurement data.

  When the sensor drift amount of measurement data used for device diagnosis is within the standard deviation of the sensor drift distribution, device diagnosis is performed using the measurement data corrected using the drift amount. Abnormality determination accuracy is improved.

  ADVANTAGE OF THE INVENTION According to this invention, the precision of the abnormality determination of the apparatus provided in the plant can be improved.

It is a block diagram of the apparatus diagnostic apparatus of Example 1 which is one preferable Example of this invention. It is explanatory drawing which shows an example of the calibration data stored in the sensor structure database shown in FIG. It is explanatory drawing which shows an example of the normal data stored in the normal database shown in FIG. It is a block diagram which shows an example of the system | strain used as the diagnostic object in the apparatus diagnostic apparatus shown in FIG. It is explanatory drawing which shows an example of the measurement data in the system | strain shown in FIG. It is a flowchart which shows the procedure of the apparatus diagnosis performed with the apparatus diagnostic apparatus shown in FIG. It is explanatory drawing which shows the other example of the calibration data stored in the sensor structure database shown in FIG. It is explanatory drawing which shows the calculation concept of the drift amount in each measurement data. It is explanatory drawing of an apparatus diagnosis. It is explanatory drawing which shows an example of an abnormality cause specific table. It is a block diagram which shows an example of the other system | system | group used as the diagnostic object in the apparatus diagnostic apparatus shown in FIG. It is a block diagram of the apparatus diagnostic apparatus of Example 2 which is another Example of this invention.

  The inventors examined a method capable of distinguishing between device abnormality and sensor drift and improving the accuracy of device diagnosis in device diagnosis performed based on the detection signal output from the sensor. Although it is difficult to previously measure data at the time of occurrence of an abnormality in a nuclear power plant, the inventors have focused on using sensor calibration data obtained during periodic inspection of the nuclear power plant. The inventors have newly found that the accuracy of equipment abnormality determination can be improved by considering the drift distribution.

  When sensors are classified according to their models and installation locations, the sensor drift amount has a characteristic according to a normal distribution. This characteristic makes it possible to estimate the average value and variance of the sensor drift at a certain time. If this drift distribution is used, it is possible to accurately evaluate the amount of drift of a sensor generated during operation of a plant such as a nuclear power plant. The device diagnosis can be performed with high accuracy by correcting the drift amount of the sensor from the measurement data and using the measurement data with the drift amount corrected.

  An embodiment of the present invention that takes the above-described knowledge into consideration will be described below.

  A device diagnostic apparatus which is a preferred embodiment of the present invention will be described with reference to FIG. The apparatus diagnostic apparatus 1 of this embodiment includes a diagnostic data input means 2, a drift amount calculation means 3, an abnormality cause identification means 4, a drift validity evaluation means (first drift determination means) 5, a drift distribution calculation means 6, a drift correction. Means 7 and device diagnosis means 8 are provided. The device diagnostic apparatus 1 has a storage device (not shown) including a sensor calibration database 9 and a normal database 10. The drift amount calculation means 3 is connected to the diagnostic data input means 2, the abnormality cause identification means 4, the drift validity evaluation means 5, the drift correction means 7, and the storage device. The drift distribution calculation means 6 is connected to the drift validity evaluation means 5 and the storage device. The drift correction unit 7 is connected to the diagnostic data input unit 2, the drift validity evaluation unit 5, and the device diagnostic unit 8. The device diagnosis unit 8 is connected to the abnormality cause identification unit 4 and the storage device. The abnormality cause identifying means 4 is connected to the display device 11.

  The sensor calibration database 9 stores calibration data, which is a record of calibration work performed at the time of periodic inspection of the nuclear power plant, for each sensor such as a flow meter, a thermometer, a pressure gauge, and a water level gauge. FIG. 2 shows an example of calibration data stored in the sensor calibration database 9. In the sensor calibration operation, an output value (an output value before calibration in FIG. 2) output from the sensor when a signal serving as a reference value shown in FIG. 2 is applied to the sensor is recorded. The difference between the output value and the reference value is a drift (drift before calibration in FIG. 2), and the sensor is calibrated so that the drift is equal to or less than a value determined for each sensor. 0%, 25%, 50%, 75%, 100%, 75%, 50%, 25%, 0% reference value obtained by adding each reference value to the sensor in order, calibration output value for the sensor, calibration The pre-drift, post-calibration output value and post-calibration drift are stored in the storage device, and the sensor calibration database 9 is updated. This calibration data is stored in the sensor calibration database 9 for each sensor provided in the nuclear power plant for each periodic inspection. If there is no sensor calibration data in a new nuclear plant that is not regularly inspected, the calibration data of the same sensor provided in the existing nuclear plant is used.

  The normal database 10 stores process values (normal data) of devices measured by the sensors when the sensors and devices are normal. This normal data is, for example, normal data measured during a start-up test and a period of several months after the start-up of the plant. FIG. 3 shows an example of process values that are normal data stored in the normal database 10. The process values 1 and 2 shown in FIG. 3 indicate changes over time in two process values, for example, flow rate, temperature, pressure, water level, rotation speed, vibration displacement, and valve opening. .

  The drift distribution calculation means 6 inputs calibration data stored in the sensor calibration database 9 and calculates the drift distribution of the sensor at an arbitrary time. Specifically, the drift distribution of the sensor at an arbitrary time is obtained based on the equations (1), (2), and (3) as described later. The diagnostic data input means 2 inputs the process value (measurement data) measured by each sensor as diagnostic data of the diagnostic object. The drift amount calculation unit 3 calculates the drift amount based on the diagnostic data input from the diagnostic data input unit 2 and the normal data input from the normal database 10. The drift validity evaluation means 5 determines whether or not the drift amount calculated by the drift amount calculation means 3 is within the standard deviation of the drift distribution obtained by the drift distribution calculation means 6.

  The drift correction means 7 inputs diagnostic data from the diagnostic data input means 2, and the drift amount calculated by the drift amount calculation means 3 in the drift validity evaluation means 5 is the standard deviation of the drift distribution obtained by the drift distribution calculation means 6. When it is determined that the data exists in the data, the corrected diagnostic data is created by subtracting the drift amount from the diagnostic data. Further, when it is determined that the drift amount calculated by the drift amount calculating unit 3 does not exist within the standard deviation of the drift distribution, the drift correction unit 7 does not correct the diagnostic data based on the drift amount, The inputted diagnostic data is output as corrected diagnostic data as it is.

  The device diagnosis unit 8 uses the correction data input from the drift correction unit 7 and the normal data input from the normal database 4 to diagnose the device that is the object of diagnosis. The abnormality cause identifying unit 4 is based on the drift amount calculated by the drift amount calculating unit 3, information on the determination result obtained by the drift validity evaluation unit 5, and information on the device diagnosis result obtained by the device diagnostic unit 8. Then, when an abnormality has occurred in the device that is the diagnostic object, the cause of the abnormality is identified. The display device 11 displays the cause of the abnormality identified by the abnormality cause identifying means 4. In addition, the display device 11 includes the drift amount obtained by the drift amount calculation unit 3, information on the determination result obtained by the drift validity evaluation unit 5, and information on the device diagnosis result obtained by the device diagnosis unit 8. It may be displayed.

  An example of a plant system which is a diagnosis target of this embodiment will be described with reference to FIG. This plant system includes an A-system cooling water supply system 20A and a B-system cooling water supply system 20B. This plant system is a part of a nuclear plant. The cooling water supply system 20A is provided with a pump 22A in a pipe 21A through which cooling water flows. A flow sensor 23A is provided in the pipe 21A downstream of the pump 22A, and a tachometer 24A is provided in the pump 22A. The cooling water supply system 20B is provided with a pump 22B in a pipe 21B through which cooling water flows. A flow sensor 23B is provided in the pipe 21B downstream of the pump 22B, and a tachometer 24B is provided in the pump 22B. The pipe 21A and the pipe 21B are branched from one pipe upstream of the pumps 22A and 22B. A control device 25 for controlling the rotational speed of the pumps 22A and 22B is provided.

  The control device 25 inputs the flow rates of the cooling water discharged from the pumps 22A and 22B measured by the flow rate sensors 23A and 23B and the rotation speeds of the pumps measured by the tachometers 24A and 24B. When the pumps 22A, 22B, the flow sensors 23A, 23B, and the tachometers 24A, 24B are normal, the respective flow rates measured by the flow sensors 23A, 23B are the same value, and the rotations measured by the tachometers 24A, 24B. The number is the same value. The control device 25 controls the rotational speed of the pump 22A based on the flow rate measured by the flow sensor 23A and the rotational speed measured by the tachometer 24A, and the flow rate measured by the flow sensor 23B and the tachometer 24B. Based on the number of revolutions measured in step 1, the number of revolutions of the pump 22B is controlled. Control of the pumps 22 </ b> A and 22 </ b> B is controlled in conjunction with the control device 25.

  The device diagnostic apparatus 1 diagnoses the state of the pump 22A, which is a device of the cooling water supply system 20A, based on the flow rate measured by the flow sensor 23A and the rotation speed measured by the tachometer 24A, and the cooling water supply system The state of the pump 22B, which is a 20B device, is diagnosed based on the flow rate measured by the flow sensor 23B and the rotational speed measured by the tachometer 24B. In sensor diagnosis of sensors, for example, the flow rate sensors 23A and 23B, the drift of the sensor is diagnosed by comparing the flow rate measured by the flow rate sensor 23A and the flow rate measured by the flow rate sensor 23B.

  The flow rate measured by the flow sensor 23A (referred to as flow rate A), the flow rate measured by the flow sensor 23B (referred to as flow rate B), the rotational speed measured by the tachometer 24A (referred to as rotational speed A), and the tachometer 24B. FIG. 5 shows an example of measurement of each rotation number (referred to as rotation number B). In these measurement examples, the flow rate B gradually decreases due to some abnormality when 100 days have elapsed from the start of operation of the plant system. In the conventional equipment diagnosis, the pump 22B is diagnosed as being abnormal because the flow rate B is decreasing while the rotational speed of the pump 22B is constant in the cooling water supply system 20B. On the other hand, in the sensor diagnosis, since the value of the flow rate B is lower than that of the flow rate A, it is diagnosed that a drift has occurred in the flow rate sensor 23B. Therefore, since both the pump 22B and the flow rate sensor 23B are diagnosed as abnormal, it is not possible to distinguish between device abnormality and sensor drift unless it is distinguished which is a wrong diagnosis.

  The procedure of device diagnosis in the present embodiment using the device diagnosis apparatus 1 will be described with reference to FIG. This equipment diagnosis was performed on the plant system shown in FIG. 5 as an example.

  The drift distribution calculation means 6 calculates the sensor drift distribution using the calibration data stored in the sensor configuration database 9 (step S1). The sensor drift has a characteristic according to a normal distribution when the sensor drift is grouped and evaluated according to the sensor of the same model and the installation location of the sensor. Calibration data for each sensor shown in FIG. 5 is stored in the sensor calibration database 9. FIG. 7 shows an example of calibration data in a sensor similar to the flow rate sensor provided in the plant system shown in FIG. The calibration data of each sensor shown in FIG. 7 indicates the drift amount at the reference value 100% shown in FIG. These drift amounts are obtained in the calibration work of the sensor in the periodic inspection of the nuclear power plant. In this example, the calibration work is performed 400 days after the previous calibration work. When the number of days until the sensor calibration is different due to the periodic inspection, the drift amount of each sensor is assumed to be proportional to the time, and is converted into, for example, 400 days. The calculation of the drift distribution in step S1 using the calibration data with the reference value of 100% will be described below as an example, but other reference values can be calculated in the same manner.

  When the sensor drift follows a normal distribution, the drift amount distribution f (x) at time t can be calculated using the equation (1).

Here, x is the drift amount, μ is the average value of the drift distribution, and σ is the standard deviation of the drift distribution.

  When the sensor 1, sensor 2, and sensor 3 of the calibration data shown in FIG. 7 are calculated using all these data as a group of the same type of sensor, the drift distribution of the reference value 100% before calibration is μ = −0. Normal distribution with .76% and σ = 0.31%. Moreover, for the sake of simplicity, all the drift amounts after calibration are approximated to 0.0%. At this time, due to the reproducibility of the normal distribution, the average value μ of the drift distribution is obtained by the equation (2), and the standard deviation σ of the drift distribution is obtained by the equation (3).

μ = μ ₀ × t / t ₀ (2)
σ = σ ₀ × (t / t ₀ ) ^0.5 (3)
Here, μ ₀ is the average value of the drift distribution at time t ₀ , and σ ₀ is the standard deviation of the drift distribution at time t ₀ . Therefore, the drift distribution represented by the equation (1) is a distribution that changes at time t, and the drift distribution at a certain time t can be calculated. Specifically, the average value and standard deviation of the drift distribution at a certain time are obtained. For example, the drift distribution at t = 200 days can be estimated as μ = −0.38% and σ = 0.22%.

  During operation of the nuclear power plant, measurement signals from the sensors provided in the nuclear power plant such as the flow sensors 23A and 23B and the tachometers 24A and 24B are input to the diagnostic data input means 2 and are converted into digital signals by A / D conversion. Is converted to These measurement signals converted into digital signals (measurement data and diagnosis data) are input from the diagnosis data input means 2 to the drift amount calculation means 3. The drift amount calculation means 3 calculates the sensor drift amount from the diagnostic data (step S2). The drift amount calculation means 3 calculates the drift amount of this sensor based on the diagnosis data input from the diagnosis data input means 2 and the normal data for the diagnosis data input from the normal database 10. The drift amount is calculated for all diagnostic data for each input sensor.

  FIG. 8 shows the relationship between the distribution of normal data of the flow rate A and the flow rate B and the diagnostic data of the flow rate A and the flow rate B at a certain pump speed of the pumps 22A and 22B. The horizontal axis in FIG. 8 indicates the flow rate A, and the vertical axis indicates the flow rate B. The flow rate A and the flow rate B that create the distribution of normal data shown in FIG. 8 have the same value. The drift amount calculation means 3 inputs normal data of the flow rate A and the flow rate B from the normal database 10, obtains the distribution of normal data shown in FIG. 8, and further, the flow rate A and the flow rate B at the center of this distribution. Find each value of. The diagnosis data of the flow rate A and the flow rate B shown in FIG. 8 is measurement data measured by the flow rate sensors 23A and 23B at a certain point in time. In the example shown in FIG. 8, each of the drift amount A of the flow rate A and the drift amount B of the flow rate B can be calculated based on the distance from the center of the distribution of normal data to the diagnostic data. The drift amount A of the flow sensor A is + 2%, and the drift amount B of the flow sensor B is −12%. The calculation of the drift amount may be performed using a known method such as a neural network.

  The drift validity evaluation means 5 evaluates the validity of the drift amount (step S3). The drift validity evaluation means 5 determines whether or not the sensor drift amount x calculated by the drift amount calculation means 3 in step S2 is within the standard deviation of the drift distribution obtained by the drift distribution calculation means 6 in step S1. To do. Specifically, if the sensor drift amount x satisfies μ−3σ <x <μ + 3σ, the drift validity evaluation means 5 determines that the calculated sensor drift amount x is within the standard deviation of the drift distribution. It is determined that it is contained, and it is evaluated that the calculated drift amount is appropriate.

  When the drift validity evaluation unit 5 determines that the sensor drift amount x is appropriate (step S4), the drift correction unit 7 corrects the diagnostic data using the sensor drift amount (step S5). The drift correction unit 7 corrects the diagnostic data by subtracting the drift amount of the sensor that has measured the diagnostic data from the diagnostic data input from the diagnostic data input unit 2. The diagnostic data corrected in this way is referred to as corrected diagnostic data. When the drift validity evaluation means 5 determines that the drift amount of the sensor does not fall within the standard deviation of the drift distribution obtained by the drift distribution calculation means 6 and the drift amount of the sensor is not valid (step S4), the drift is evaluated. The correcting means 7 outputs the inputted diagnostic data as it is as corrected diagnostic data without correcting the diagnostic data inputted from the diagnostic data input means 2 with the drift amount.

  The device diagnosis unit 8 performs device diagnosis using the corrected diagnosis data (step S6). The device diagnosis unit 8 inputs correction diagnosis data from the drift correction unit 7. When the drift validity evaluation means 5 determines that the drift amount of the sensor calculated for certain diagnostic data is appropriate, the device diagnostic means 8 uses the corrected diagnostic data that substantially corrects this diagnostic data, When it is determined by the drift validity evaluation means 5 that the calculated drift amount is not valid, the target device is diagnosed using the corrected diagnostic data that remains as the diagnostic data.

  The distribution of past normal data of the flow rate B measured by the flow rate sensor 23B and the rotation speed B measured by the tachometer 24B in the cooling water supply system 20B and the diagnostic data measured at a certain time are shown in FIG. Suppose that it is in a state. The two state quantities (flow rate and rotation speed) that create the distribution of normal data shown in FIG. 9 are proportional to each other. The device diagnosis means 8 inputs normal data of the flow rate B and the rotational speed B stored from the normal database 10, obtains the distribution of the normal data of the flow rate B and the rotational speed B shown in FIG. The values of the flow rate B and the rotation speed B at the center of this distribution are obtained. The diagnosis data of the flow rate B and the rotation speed B shown in FIG. 9 are measurement data measured by the flow rate sensor 23B and the tachometer 24B at a certain point in time. For example, when the distance (abnormality degree) from the center of the distribution of normal data to the diagnostic data exceeds a threshold value, the device diagnosis unit 8 diagnoses that there is an abnormality. In the device diagnosis performed by the device diagnosis means 8, a known method such as clustering may be used.

  The abnormality cause identifying means 4 identifies the cause of the abnormality (step S7). The abnormality cause identifying unit 4 determines the drift amount obtained by the drift amount calculating unit 3 (the drift amount obtained by the processing of step S2) and the determination information on the validity of the drift amount obtained by the drift validity evaluating unit 5 ( When the determination information obtained by the processing of steps S3 and S4) and the information of the device diagnosis obtained by the device diagnosis means 8 (device diagnosis information obtained by the processing of step S6) are input, and an abnormality has occurred. In addition, the cause of the abnormality is specified based on the input information. The abnormality cause identification unit 4 stores the abnormality cause identification table shown in FIG. 10, and uses information of the abnormality cause identification table when identifying the abnormality cause.

  When the drift validity evaluation means 5 determines that the drift amount is appropriate, the drift amount exists within the range of the drift distribution, that is, within the standard deviation of the drift distribution, and the diagnostic data is corrected using the drift amount. In addition, since the device diagnosis means 8 performs device diagnosis using the corrected diagnosis data, the result of the device diagnosis by the device diagnosis means 8 is not affected by drift. Therefore, when the drift validity evaluation unit 5 determines that the drift amount is appropriate, the abnormality cause identification unit 4 determines that the device is normal when the device diagnosis unit 8 diagnoses that the device to be diagnosed is normal. When the device diagnosis means 8 diagnoses that the device to be diagnosed is abnormal, it is determined that the device that is abnormal is abnormal.

  When the drift validity evaluation unit 5 determines that the drift amount is appropriate, the abnormality cause identification unit 4 further determines whether the drift amount obtained by the drift amount calculation unit 3 is less than or equal to the drift allowable value. . If the drift amount of the sensor is larger than the allowable drift value, the sensor that has measured the diagnostic data is in a state that needs to be calibrated. Judge that there is. The determination of “sensor drift” by the abnormality cause identifying unit 4 is performed when any of the determination results of “device normal” and “device abnormality” is given when the drift amount is appropriate. If the drift amount of the sensor is equal to or less than the allowable drift value, it is not determined that the sensor is “sensor drift” because calibration of the sensor is unnecessary.

  The abnormality cause identifying process in the abnormality cause identifying means 4 when the drift validity evaluation means 5 determines that the drift amount is not within the standard deviation of the drift distribution and the drift amount is not valid will be described.

  The abnormality cause identifying means 4 is when the drift amount is outside the drift distribution (the drift amount is outside the standard deviation of the drift distribution), and the device diagnosis means 8 diagnoses that the device to be diagnosed is abnormal Is determined to be caused by the device diagnosed by the device diagnosis means 8. For this reason, the abnormality diagnosed by the sensor diagnosis is a misdiagnosis affected by the device abnormality. Therefore, the abnormality cause identification unit 4 determines that the diagnosis target device is abnormal.

  The abnormality cause identifying unit 4 has a case where the drift amount is outside the drift distribution (the drift amount is outside the standard deviation of the drift distribution) and the device diagnosis unit 8 diagnoses that the device to be diagnosed is normal. Is determined to be other abnormality. This is because a certain degree of abnormality has occurred in the device to be diagnosed, and due to this, the sensor is diagnosed as having a sensor abnormality, but the degree of abnormality of this device is below the threshold in the device diagnosis. This is because it is diagnosed as normal by device diagnosis.

  The abnormality cause identification table shown in FIG. 10 is an example, and an abnormality cause identification table classified in more detail using the drift amount of the sensor and the abnormality level of the device diagnosis may be used.

  Each abnormality cause identification information obtained by the abnormality cause identification means 4 is output to the display device 11 (step S8) and displayed on the display device 11. When the drift amount is within the standard deviation of the drift distribution (the drift amount is within the drift distribution), the displayed error cause identification information is “device normal”, “device normal, sensor drift”. , "Equipment abnormality" and "Equipment abnormality, sensor drift", and when the drift amount is outside the standard deviation of the drift distribution (the drift amount is outside the drift distribution) "Abnormality" or "equipment abnormality".

  A specific example of device diagnosis in the device diagnosis apparatus 1 described above will be described using measurement data (diagnosis data) of each sensor shown in FIG. In the measurement result shown in FIG. 5, −10% drift occurs in the measurement value of the flow rate sensor 23B on the 200th day. For example, it is assumed that the drift distribution of the flow rate sensor 23B on the 200th day based on the calibration data is μ = −7% and σ = 2%. Since the drift amount calculated in step S2 exists in the drift distribution, the measurement data (diagnosis data) of the flow sensor 23B is corrected using the drift amount in step S5. It is determined to be normal in device diagnosis (step S6) using the corrected diagnosis data. In the abnormality cause identifying process in step S7, it is determined that the sensor drift is the cause. If μ is −1% and σ is 0.5%, the drift amount exists outside the drift distribution, so the diagnostic data is not corrected using the drift amount. Since the device diagnosis using the uncorrected diagnosis data (step S6) is determined to be abnormal, it is determined that the device abnormality is the cause (step S7).

  In this embodiment, when the drift amount of measurement data used for device diagnosis of the sensor is within the standard deviation of the drift distribution of the sensor, the device diagnosis is performed using the measurement data corrected using the drift amount. As a result, the accuracy of device abnormality determination is improved. Conventionally, it is possible to accurately determine an abnormality of a device that cannot be diagnosed due to the effect of sensor drift. In addition, when the drift amount of measurement data used for sensor diagnosis of the sensor is outside the standard deviation of the drift distribution of the sensor, device diagnosis is performed using measurement data that is not corrected using the drift amount. Even when the drift amount is outside the standard deviation of the drift distribution of the sensor, the accuracy of device diagnosis is improved.

  In the present embodiment, when the drift amount of the sensor used for diagnosing the sensor is within the standard deviation of the drift distribution of the sensor, the drift amount is less than the drift allowable value in the abnormality cause identifying means 4. Therefore, it is possible to determine that a sensor drift has occurred separately from the determination of whether the device is abnormal or normal.

  The nuclear power plant includes, for example, the system shown in FIG. In this system, a pump 22 for boosting the cooling water is provided in a pipe 21 through which the cooling water flows. A tachometer 24 is installed in the pump 22, a flow sensor 23 </ b> C is provided in the pipe 21 upstream of the pump 22, and a flow sensor 23 </ b> D is provided in the pipe 21 downstream of the pump 22. The flow rate sensor 23C and the flow rate sensor 23D have the same flow rate measurement data for each sensor when normal.

  The apparatus diagnosis apparatus 1 according to the present embodiment performs the abnormality diagnosis of the pump 22 that is an apparatus as described above according to the processing procedure illustrated in FIG. 6 similarly to the system illustrated in FIG. 4 even in the system illustrated in FIG. It can be carried out. As shown in FIG. 8, the respective drift amounts of the flow rate C measured by the flow rate sensor 23C and the flow rate D measured by the flow rate sensor 23D are respectively normal data of the flow rate C and the flow rate D, and the flow rate sensors 23C and 23D. It can obtain | require using the measured measurement data (diagnosis data).

  As shown in FIG. 9, the diagnosis of the pump 22 which is executed by the device diagnosis means 8 is performed as follows: the distribution of each normal data of the measurement data of the tachometer 24 and the measurement data of the flow sensor 23D, 24, and measurement data of the flow rate sensor 23D. Each effect mentioned above can be acquired also by diagnosis of the apparatus provided in the system | strain shown in FIG.

  An apparatus diagnosis apparatus according to another embodiment of the present invention will be described with reference to FIG. The device diagnostic apparatus 1A of the present embodiment is a software implementation of the device diagnostic apparatus 1 of the first embodiment using a computer.

  The apparatus diagnosis apparatus 1A according to the present embodiment includes a computer that is an information processing apparatus, and includes a central processing unit (CPU) 26, a memory 27, an input / output interface 28, and an input device 29. The central processing unit (CPU) 26, the memory 27, the input / output interface 28, and the input device 29 are connected to each other through an internal run of the device diagnostic apparatus 1A. Sensors such as flow sensors 23A and 23B and tachometers 24A and 24B provided in the nuclear power plant, and the display device 30 are connected to the input / output interface 28.

  During operation of the nuclear power plant, measurement data measured by each sensor is input from the input / output interface 28 to the device diagnostic apparatus 1A and stored in the memory 27. A program including the processing procedure of steps S1 to S8 shown in FIG. The memory 27 includes a sensor calibration database 9 and a normal database 10. The CPU 26 executes device diagnosis in the same manner as in the first embodiment, based on a program including the processing procedure of steps S1 to S8 read from the memory 27. That is, the CPU 26 sequentially executes the processes of steps S1 to S8 and executes the device diagnosis performed in the first embodiment.

  Also in this embodiment, each effect produced in the first embodiment can be obtained.

  Examples 1 and 2 can be applied not only to a nuclear power plant but also to diagnosis of equipment provided in other plants such as a thermal power plant and a chemical plant.

  The relationship between the sensors for determining the sensor drift amount (for example, the flow rate sensor 23A and the flow rate sensor 23B shown in FIG. 4 and the flow rate sensor 23C and the flow rate sensor 23D shown in FIG. 11) is correlated with the state quantities measured by each other. Have a changing relationship. Specifically, both sensors have a state quantity measured by one sensor showing the same value as the state quantity measured by the other sensor, or the former state quantity is proportional to the latter state quantity. This relationship is such that the state quantity measured by one sensor can be estimated by the state quantity measured by the other sensor.

  The present invention can be used for diagnosis of equipment provided in a plant such as a nuclear power plant.

  DESCRIPTION OF SYMBOLS 1,1A ... Apparatus diagnostic apparatus, 2 ... Drift distribution calculation means, 3 ... Drift amount calculation means, 4 ... Abnormal cause identification means, 5 ... Drift validity evaluation means, 6 ... Drift distribution calculation means, 7 ... Drift correction means, DESCRIPTION OF SYMBOLS 8 ... Device diagnostic means, 9 ... Sensor calibration database, 10 ... Normal database, 26 ... Central processing unit (CPU), 27 ... Memory, 28 ... Input-output interface.

Claims (6)

In a device diagnosis method for diagnosing a device provided in a plant using measurement data measured by a plurality of sensors provided in the plant,
A drift amount of measurement data measured by the sensor is calculated, a drift distribution of the sensor is obtained based on a plurality of calibration data of the sensor, and the calculated drift amount is within a standard deviation of the drift distribution. And when the calculated drift amount is determined to be within the standard deviation of the drift distribution, the measurement data of the sensor that has output the measurement data used for the calculation of the drift amount, A device diagnosis method characterized by correcting using a drift amount and diagnosing the device based on the corrected measurement data.   When it is determined that the calculated drift amount is outside the standard deviation of the drift distribution, the measurement data used for the calculation of the drift amount is not corrected using the drift amount, and this measurement data is not corrected. The apparatus diagnosis method according to claim 1, wherein diagnosis of the apparatus is performed using a device.   The device diagnosis method according to claim 1, wherein when it is determined that the calculated drift amount is within a standard deviation of the drift distribution, it is determined whether the drift amount is equal to or less than a drift allowable value.   Drift amount calculating means for calculating a drift amount of measurement data measured by a sensor provided in a plant; drift distribution calculating means for obtaining a drift distribution of the sensor based on a plurality of calibration data of the sensor; First drift determination means for determining whether the drift amount is within the standard deviation of the drift distribution, and when it is determined that the calculated drift amount is within the standard deviation of the drift distribution, the drift amount Drift correction means for correcting the measurement data of the sensor that has output the measurement data used for the calculation using the drift amount, and diagnosis of equipment provided in the plant based on the corrected measurement data An apparatus diagnosis apparatus comprising an apparatus diagnosis unit for performing the apparatus diagnosis. When the drift correction unit determines that the calculated drift amount is outside the standard deviation of the drift distribution, the measurement data of the sensor that has output the measurement data used for the calculation of the drift amount, Drift correction means for outputting as corrected measurement data without correcting using the drift amount,
5. The apparatus diagnosis apparatus according to claim 4, wherein the apparatus diagnosis means is an apparatus diagnosis means for diagnosing the apparatus based on the corrected measurement data when the corrected measurement data is input.   The second-degree lift determination means for determining whether or not the drift amount is equal to or less than a drift allowable value when it is determined that the calculated drift amount is within a standard deviation of the drift distribution. The apparatus diagnostic apparatus as described in.

Images