JP6347771B2 - Abnormality diagnosis apparatus, abnormality diagnosis method, and abnormality diagnosis program - Google Patents

Description

  The present invention relates to an abnormality diagnosis device, an abnormality diagnosis method, and an abnormality diagnosis program for diagnosing plant abnormality.

  In the plant, a plurality of devices are operated under various conditions. In the plant, the state is measured using a plurality of sensors. An abnormality of the plant can be detected from the plurality of condition values and the measurement values of the plurality of sensors. For example, the upper and lower limit values (threshold values) determined for the measurement values of the sensors by the measurement values of the sensors are set. The mainstream method is to determine that there is an abnormality when the number is exceeded. On the other hand, when the upper and lower limit values are set for each measured value of each sensor, it is necessary to increase the number of upper and lower limit values when the number of sensors to be used is increased.

  On the other hand, in recent years, there is also a technique for diagnosing abnormality using a pattern recognition technique called MT method (Mahalanobis Taguchi method). The MT method is for diagnosing whether or not the state is “same as usual”, and the Mahalanobis distance of the object to be diagnosed is determined from the Mahalanobis space based on the center of the Mahalanobis space that represents the same state as usual. This is a method for determining an abnormality when going far. Specifically, there is one that divides sensor measurement data into a plurality of groups, generates a unit space for each group, and diagnoses an abnormality using the MT method (see, for example, Patent Document 1).

Here, even if it is normal, the state varies depending on the driving load, season, time, or the like. Therefore, it is desirable to make a diagnosis using a unit space that matches the state of the plant at the time of diagnosis. On the other hand, when the state is automatically changed by automatic control, it is difficult to generate a unit space by judging the state of the plant in real time.

JP 2010-181188 A

  As described above, it is difficult to diagnose an abnormality with high accuracy according to the real-time state of the plant.

  In view of the above problems, an object of the present invention is to provide an abnormality diagnosis device, an abnormality diagnosis method, and an abnormality diagnosis program that perform an abnormality diagnosis with high accuracy according to a real-time state of a plant.

  In order to achieve the above object, the present invention provides an abnormality diagnosis device for diagnosing plant abnormality, a group data storage unit in which a group including records including variables representing the operation state of a plant is stored, and a predetermined data For each of the plurality of groups read from the group data storage unit repeatedly at the timing, the value of the variable included in the group and the value of the variable representing the new state of the plant corresponding to the variable are used. A first calculation unit for calculating the Mahalanobis distance, a selection unit for selecting the group having the smallest Mahalanobis distance from the plurality of groups, and reading from the group data storage unit at a predetermined timing for diagnosing abnormality. For the group selected by the selected selection unit, the value of the variable included in the group and the previous value corresponding to the variable A second calculation unit that calculates a Mahalanobis distance using a value of a variable that represents a new state of the plant, and a diagnosis unit that determines the state of the plant using the Mahalanobis distance calculated by the second calculation unit; It is characterized by providing.

  The present invention also relates to an abnormality diagnosis method performed by an abnormality diagnosis apparatus for diagnosing plant abnormality, wherein a predetermined timing is obtained from a group data storage unit in which a group including records including variables representing the operation state of a plant is stored. First, the Mahalanobis distance is calculated for each of the plurality of groups read repeatedly in the above using the value of the variable included in the group and the value of the variable representing the new state of the plant corresponding to the variable. 1 calculation step, a selection step of selecting the group having the smallest Mahalanobis distance among the plurality of groups, and reading from the group data storage unit at a predetermined timing for diagnosing an abnormality and selecting in the selection step The value of the variable included in the group and the plant corresponding to the variable A second calculation step of calculating a Mahalanobis distance using a value of a variable representing a new state, and a diagnosis step of determining a plant state using the Mahalanobis distance calculated in the second calculation step; It is characterized by providing.

  Furthermore, the present invention is an abnormality diagnosis program that causes a computer to function as the abnormality diagnosis apparatus.

  According to the present invention, the accuracy of plant abnormality diagnosis can be improved.

It is a block diagram explaining the abnormality diagnosis apparatus concerning an embodiment. It is a data block diagram explaining an example of the accumulation | storage data utilized with an abnormality diagnosis apparatus. It is a flowchart explaining an example of the process in an abnormality diagnosis apparatus. It is a conceptual diagram explaining the abnormality diagnosis using MT method.

  The abnormality diagnosis apparatus according to the embodiment is an abnormality diagnosis apparatus that diagnoses an abnormality in plant operation. For example, the plant diagnosed by the abnormality diagnosis device is a power plant. The power plant includes a plurality of devices (pumps, valves, etc.), and values for controlling these devices are set as target values. This target value is, for example, the pressure of the pump, the opening / closing of a valve, or the like. In addition, the power plant includes a plurality of sensors, and each sensor measures the intake air amount, the exhaust air amount, the power generation amount, and the like.

  Furthermore, in a power plant, various modes can be set, and the target value set for each device and the measured value measured by each sensor differ depending on this mode. For example, in a power plant, it is not always necessary to generate a certain amount of power, and power is generated in accordance with power demand. Therefore, a plurality of modes are defined according to the time zone, day of the week, season, etc., and the target value of each device differs for each set mode. Moreover, since the target value of each device is different, the measured value of each sensor is different for each set mode.

  As shown in FIG. 1, the abnormality diagnosis apparatus 1 according to the embodiment generates an update unit 11 that updates accumulated data 21 with new variables input from a plant, and groups used for abnormality diagnosis. A generation unit 12 that sets determination values used in diagnosis and stores them as group data 22; a first calculation that calculates Mahalanobis distances for a plurality of groups using variables newly input from the plant and variables of group data 22 Unit 13, selection unit 14 for selecting the minimum group among Mahalanobis distances calculated by first calculation unit 13, variables newly input from the plant, and variables of group data 22 of the group selected by selection unit 14 The second calculation unit 15 that calculates the Mahalanobis distance using the value and the Mahalanobis distance calculated by the second calculation unit 15 is compared with the determination value. And a diagnosis unit 16 for diagnosing the normal.

  The abnormality diagnosis apparatus 1 is, for example, an information processing apparatus including a central processing unit (CPU) 10 and a storage device 20, and the abnormality diagnosis program P stored in the storage device 20 is read and executed. As illustrated in FIG. 1, an update unit 11, a generation unit 12, a first calculation unit 13, a selection unit 14, a second calculation unit 15, and a diagnosis unit 16 are mounted on the CPU 10. In addition to the abnormality diagnosis program P, the storage device 20 stores accumulated data 21 and group data 22. The abnormality diagnosis device 1 is connected to an input device 2 such as a keyboard, a mouse, an operation button, and a touch panel for inputting operations, and an output device 3 such as a display and a speaker for outputting an abnormality diagnosis process and results. It is connected.

  When the value of each variable is newly input from the plant, the update unit 11 generates a new record including the value of each input variable, adds the generated record, and stores the accumulated data 21 stored in the storage device 20. Update.

  The accumulated data 21 is data obtained by accumulating variable records representing past plant states. The variables of the accumulated data 21 are the plant operation condition values and the measured values measured in the plant in the case of the condition values. In the example shown in FIG. 2, the accumulated data 21 includes a target value (variables 1 and 2) set in each device of the plant, which is a condition value for plant operation, and a measured value measured in the case of this mode and the target value. (Variables 4, 5) and the like are associated with the record. In the example of the accumulated data 21 shown in FIG. 2, each record is given a sample number such as time.

  The generation unit 12 reads the accumulated data 21 from the storage device 20 at a predetermined timing for generating a group, extracts records in which variables satisfy a predetermined extraction condition, and generates a plurality of groups. This extraction condition is input via the input device 2, for example. For example, any one of the values of variable 1 and variable 2 (target value), variable 4 and variable 5 (measured value) may be used as the extraction condition, or a combination of values of a plurality of variables may be used as the extraction condition. good. The generation unit 12 sets determination values used for plant abnormality diagnosis in the plurality of generated groups, generates group data 22 including the values and determination values of the variables in the group, and stores them in the storage device 20.

  In the example of FIG. 1, the storage device 20 is shown as storing one group data 22, but the storage device 20 stores group data 22 for each group.

  Here, the predetermined timing at which the generation unit 12 generates a group is, for example, a periodic timing or a timing input via the input device 2. Moreover, the value used as the determination value by the generation unit 12 is a value representing the Mahalanobis space obtained for this group. This determination value is input via the input device 2, for example.

  The first calculation unit 13 reads the group data 22 of each group from the storage device 20 at a predetermined timing for selecting a group to be used for abnormality diagnosis, and extracts the value of each variable included in the read group data 22. In addition, the first calculation unit 13 obtains the Mahalanobis distance of each group by using the value of each variable extracted from the group data 22 and the new variable input from the plant at a predetermined timing for diagnosing abnormality. The obtained value is output to the selection unit 14. Here, the predetermined timing for selecting a group to be used for abnormality diagnosis is, for example, a periodic timing or a timing at which an abnormality is diagnosed by the diagnosis unit 16.

  Specifically, (1) First, the first calculation unit 13 obtains an average value and a standard deviation for each variable. (2) Then, the 1st calculation part 13 standardizes data using the value of each variable, the average value calculated | required with respect to each variable, and a standard deviation, and calculates | requires the normalized value with respect to each variable. (3) Subsequently, the first calculation unit 13 obtains a correlation matrix for each variable by using the normalized value obtained for each variable, and obtains an inverse matrix of the correlation matrix. (4) Finally, the first calculator 13 obtains the Mahalanobis distance using the obtained inverse matrix. Based on the Mahalanobis distance determined in this way, it is possible to identify the difference between the current state and the normal state for the value of the target group, and to determine the difference between the current state and the normal state in the plant. it can.

  When the selection unit 14 inputs the Mahalanobis distance of each group from the first calculation unit 13, the selection unit 14 selects a minimum value from the plurality of input Mahalanobis distances, and selects the group having the minimum value as a group to be used for abnormality diagnosis.

  The second calculation unit 15 reads the group data 22 of the group used for diagnosis of the abnormality selected by the selection unit 14 from the storage device 20 at a predetermined timing for diagnosing the abnormality, and each variable included in the read group data 22 Extract the value of. In addition, the second calculation unit 15 obtains the Mahalanobis distance using the values of each variable extracted from the group data 22 and the new variable input from the plant at a predetermined timing for diagnosing the abnormality, and sends the Mahalanobis distance to the diagnosis unit 16. Output the calculated value. The calculation method of the Mahalanobis distance in the second calculation unit 15 is the same as the calculation method in the first calculation unit 13.

  Here, the predetermined timing for diagnosing abnormality is, for example, a regular timing. In addition, the second calculation unit 15 calculates the Mahalanobis distance at a shorter time interval than the first calculation unit 13. That is, in the abnormality diagnosis device 1, the plurality of Mahalanobis distances calculated by the first calculation unit 13 are used, and the selection unit 14 selects a group used for abnormality diagnosis. Thereafter, the second calculation unit 15 continues the calculation of the Mahalanobis distance for diagnosing plant abnormality using the group selected by the selection unit 14, but the second calculation unit 15 selects the group according to the state of the plant. 1 The Mahalanobis distance is calculated again at the timing when the calculation unit 13 confirms the group. Therefore, the second calculation unit 15 calculates the Mahalanobis distance with a shorter time span than the first calculation unit 13.

  When the first calculation unit 13 calculates a plurality of Mahalanobis distances again, when it is desirable to change the group used for diagnosis of abnormality because the state of the plant has changed, the selection unit 14 In order to select a different group, the second calculation unit 15 then calculates the Mahalanobis distance of the newly selected group. On the other hand, when there is no need to change the group used for the diagnosis of abnormality, the selection unit 14 selects the same group as the previous time, so that the second calculation unit 15 continues the Mahalanobis of the same group as the previous time. Calculate the distance.

  When the Mahalanobis distance is input from the second calculation unit 15, the diagnosis unit 16 reads the group data 22 used for abnormality diagnosis from the group data 22, and extracts a determination value included in the read group data 22. In addition, the diagnosis unit 16 compares the Mahalanobis distance input from the second calculation unit 15 with the determination value extracted from the group data 22, and determines whether the operation state of the plant is normal or abnormal. Is output to the output device 3.

  For example, as shown in FIG. 4, when determining an abnormality using the MT method, when the Mahalanobis distance obtained by the second calculation unit 15 is smaller than a set determination value, that is, within the Mahalanobis space, diagnosis is performed. The unit 16 determines that the current plant state is the same as the normal state, and diagnoses that the plant is operating normally. On the other hand, when the Mahalanobis distance obtained by the second calculating unit 15 is larger than the set determination value, that is, when the Mahalanobis space is out of the Mahalanobis space, the diagnosis unit 16 is in a state where the current plant state is different from the usual state. It is diagnosed that an abnormality has occurred in the plant.

  The first calculation unit 13 calculates the Mahalanobis distance of each group in a regular time span longer than the second calculation unit 15 as described above, and also detects an abnormality when the diagnosis unit 16 diagnoses an abnormality. You may calculate the Mahalanobis distance of each group as a timing which selects the group to diagnose. That is, even when the Mahalanobis distance of the group calculated by the second calculation unit 15 is outside the range of the predetermined determination value, the diagnosis in the group that has been used is not optimal due to the change in the state of the plant. It may have disappeared. In this case, diagnosing an abnormality in the group that has been used has low reliability in the accuracy of the abnormality diagnosis. Therefore, even if the first calculation unit 13 calculates the Mahalanobis distance of each group again, the accuracy of the abnormality diagnosis can be relied upon when the used group is selected and an abnormality is diagnosed. .

  Moreover, the abnormality diagnosis apparatus 1 may be comprised from several information processing apparatus, for example, only the update part 11 may be contained in the information processing apparatus different from the other process parts 12-16. Further, only a part of the data stored in the storage device 20 may be stored in an external storage device.

  Next, a process for diagnosing an abnormality in an optimum group in the abnormality diagnosis apparatus 1 will be described using the flowchart shown in FIG.

  The generation unit 12 reads the accumulated data 21 from the storage device 20 at the timing when the group data 22 is generated, extracts a record value satisfying the extraction condition, generates the group data 22 and stores the group data 22 in the storage device 20 (S1). ).

  Thereafter, the first calculator 13 calculates the Mahalanobis distance for each group (S2). Subsequently, the selection unit 14 selects a minimum value from the plurality of Mahalanobis distances calculated by the first calculation unit 13, and selects a group having the minimum value of the Mahalanobis distance as a group to be used for abnormality diagnosis (S3).

  The second calculation unit 15 obtains the Mahalanobis distance using the value of the variable newly input from the plant and the value of each variable of the group data 22 for the group selected in step S3 at the timing of diagnosing the abnormality. To the diagnosis unit 16 (S4).

  The diagnosis unit 16 that has input the Mahalanobis distance from the second calculation unit 15 compares the input Mahalanobis distance with the determination value set in the group used for diagnosis of abnormality in the group data 22, and the operation state of the plant is abnormal. A diagnosis is made and the diagnosis result is output to the output device 3 (S5).

  Thereafter, the abnormality diagnosis apparatus 1 repeats the Mahalanobis distance calculation and abnormality diagnosis using the group data 22 selected by the selection unit 14 until the group confirmation timing comes (NO in S6) (S4 and S5). . Further, in the abnormality diagnosis device 1, when it is time to confirm the group (YES in S6), the process returns to step S2 and repeats the processes of steps S2 to S5.

  In addition, although it does not show about repeating the production | generation of the group data 22 in the flowchart shown in FIG. 3, the group data 22 may be repeatedly produced | generated at predetermined timings, such as regularly, for example.

  As described above, the abnormality diagnosis apparatus according to the present invention calculates the Mahalanobis distance using group data selected according to the real-time state of the plant and diagnoses the abnormality, so that the abnormality diagnosis can be performed with high accuracy. Can do.

  As mentioned above, although this invention was demonstrated in detail using embodiment, this invention is not limited to embodiment described in this specification. The scope of the present invention is determined by the description of the claims and the scope equivalent to the description of the claims.

1 ... Abnormality diagnosis device 10 ... CPU
DESCRIPTION OF SYMBOLS 11 ... Update part 12 ... Generation part 13 ... 1st calculation part 14 ... Selection part 15 ... 2nd calculation part 16 ... Diagnosis part 20 ... Storage device (accumulation data storage part, group data storage part)
21 ... Accumulated data 22 ... Group data P ... Abnormality diagnosis program 2 ... Input device 3 ... Output device

Claims (4)

An abnormality diagnosis device for diagnosing plant abnormality,
A group data storage unit for storing a group of records including variables representing the operating state of the plant;
For each of the plurality of groups read from the group data storage unit at a predetermined timing for selecting a group to be used for abnormality diagnosis, the value of a variable included in the group and a new value of the plant corresponding to the variable A first calculation unit that calculates a Mahalanobis distance using a value of a variable that represents an uncertain state;
A selection unit for selecting a group having the smallest Mahalanobis distance among the plurality of groups;
For a group selected by the selection unit read from the group data storage unit at a predetermined timing for diagnosing abnormality, a value of a variable included in the group and a new state of the plant corresponding to the variable are displayed. A second calculation unit that calculates the Mahalanobis distance using the value of the variable to represent,
An abnormality diagnosis device comprising: a diagnosis unit that determines a state of the plant using the Mahalanobis distance calculated by the second calculation unit.   The abnormality diagnosis apparatus according to claim 1, wherein the second calculation unit calculates the Mahalanobis distance at a shorter time interval than the first calculation unit. An abnormality diagnosis method performed by an abnormality diagnosis apparatus for diagnosing plant abnormality,
At a predetermined timing for selecting a group to be used for abnormality diagnosis, a plurality of groups are read from a group data storage unit in which a group including a record including a variable representing a plant operating state is stored, and For each, a first calculation step of calculating a Mahalanobis distance using a value of a variable included in the group and a value of a variable representing a new state of the plant corresponding to the variable;
A selection step of selecting a group having the smallest Mahalanobis distance among the plurality of groups;
For a group read from the group data storage unit and selected in the selection step at a predetermined timing for diagnosing an abnormality, a value of a variable included in the group and a new state of the plant corresponding to the variable are displayed. A second calculation step of calculating the Mahalanobis distance using the value of the variable to represent;
A diagnostic step of determining a state of the plant using the Mahalanobis distance calculated in the second calculation step.   An abnormality diagnosis program for causing a computer to function as the abnormality diagnosis apparatus according to claim 1.

Images