JPH0674873A - Abnormality diagnosis device - Google Patents

Abstract

PURPOSE:To improve a simulation itself by enhancing the initial accuracy of an initial value of a simulation wave which is compared with an actual amount of process. CONSTITUTION:The amount of process from a plant 1 is converted to digital values at each sampling period by a data collector 2, is collected, and then is transmitted to a cause candidate selector 3 and a recorder 5. When an abnormal change in the amount of process is detected by the candidate selector 3, a plurality of possible cause candidates are generated, the amount of process is simulated in inverse direction regarding time for each cause candidate with the current time and the amount of process as initial values by an inverse simulator 4, the simulation waveform for each cause candidate which is the above result is compared with the recorded value of the amount of actual process in the past which is recorded by the recorder 5 using a cause identification device 6, thus identifying the cause of abnormality and display it on a display 7.

Description

Detailed Description of the Invention

[0001]

The present invention relates to nuclear power, thermal power, chemistry,
The present invention relates to an abnormality diagnosis device applied to identify the cause of abnormality in various plants such as pumps and cold heat.

[0002]

2. Description of the Related Art In the conventional abnormality diagnosis technique, when identifying the cause of the abnormality by applying the simulation technique, it is usual to execute a forward simulation with respect to time. FIG. 2B is a conceptual diagram of an abnormality cause identification method in a conventional abnormality diagnosis device. In the figure, the vertical axis represents the process amount x, and the horizontal axis represents the time t.

When an abnormality occurs at the time to shown in the figure, the process amount x after to f changes due to the action of the cause of the abnormality. After that, when trying to identify the cause of the abnormality at the current time t1, the assumed causes are p, q, r
And In the conventional technique, the process value xo at the time of occurrence of abnormality to is used as an initial value, and the simulation is executed from time to to tf using the following simulation model. The flowchart of this simulation is shown in FIG.
Shown in. dx / dt = f (x, u (t)) (1) where u (t) is the disturbance input to the plant.
The content of the function f differs as follows according to the type of cause.

[0004]

[Equation 1]

Therefore, the result of the simulation differs according to the assumed cause type (p, q, r). This result is shown by the waveforms xp and x shown in FIG.
q and xr. The simulation waveform xp for which the true cause p is set matches the record of the actual process amount x, and
Simulation waveforms xq and xr assuming causes q and r
Does not match the record of the actual process amount x. Therefore, the cause p is identified as the true abnormal cause. The above is a typical example of the conventional technique for abnormality diagnosis.

[0006]

In the above prior art, it was necessary to accurately measure the time point t o of the abnormality and the process amount x o at that time. However, it is actually difficult to specify the time point t o at which the abnormality occurs. This is because the abnormality is detected by detecting the change in the process amount x. In other words, since the abnormality is detected after a corresponding time has elapsed from the true abnormality occurrence time to and the process amount x has changed, the abnormality occurrence time to cannot be specified accurately.

Therefore, it is difficult to accurately set the true abnormality occurrence time point to and the process amount xo at that time as initial values in the simulation executed in the forward direction with respect to time, and therefore the accuracy of the simulation deteriorates. As a result, there is a problem that it is difficult to confirm the agreement between the actual process amount recording and the simulation waveform comparison.

The present invention has been made in view of the above circumstances, and an object thereof is to improve the accuracy of setting an initial value of a simulation waveform to be compared with an actual waveform, which is a record of an actual process amount, so that the simulation itself. An object of the present invention is to provide an abnormality diagnosis device with high accuracy.

[0009]

According to the present invention, in an abnormality diagnosing apparatus for diagnosing an abnormality in a process of various plants, a current time and a process amount at that time are set as initial values for simulation, Based on the result of this comparison, the simulator that executes the simulation in the opposite direction with respect to each of the multiple possible causes of abnormality is compared with the obtained simulation waveform and the actual waveform obtained from the past recorded values of the actual process amount. And a cause identifying device for identifying the cause of the abnormality.

[0010]

According to the present invention, the simulation is executed in the reverse direction with respect to time based on the flowchart shown in FIG. At this time, the initial value is the value t = t1, at the present time (at the time of starting the simulation) shown in FIG.
Since x = x1, it is accurately measured and set. In this way, since the initial value of the simulation is accurately set and measured, the calculation accuracy of the simulation waveform is high, and there is no problem in comparison with the recorded waveform of the actual process amount. Therefore, it is possible to surely evaluate whether or not the two waveforms match.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of an abnormality diagnosis device in this embodiment. In the figure, a plant 1 is connected to a data collection device 2 via a transmission line 8.

The data collection device 2 is for converting the process amount input from the plant 1 into digital data at every sampling time interval and collecting it, and is connected to the cause candidate selection device 3 via the transmission line 9. . Further, the data collection device 2 is
It is also connected to the recording device 5 via a transmission line 10.

The cause candidate selecting device 3 is a data collecting device 2
When an abnormal change in the process amount output from the sensor is detected, a plurality of possible causes (cause control) are generated, and the cause is connected to the inverse simulator 4 via the transmission line 11.

The inverse simulator 4 is a cause candidate selection device 3
Based on the cause that is sequentially sent from the
It is connected to the cause identifying device 6 via a transmission line 12.

The recording device 5 records the time sent from the data collecting device 2 and the value of the process amount corresponding to the time, and the cause identifying device 6 is provided via the transmission line 13.
It is connected to the.

The cause identifying device 6 compares the simulated waveform, which is the result of the simulation sent from the inverse simulator 4, with the actual waveform, which is the record of the actual process amount sent from the recording device 5, to determine the abnormality. Identify the cause. The cause identifying device 6 is connected to the display device 7 via the transmission line 14. The display device 7 displays the cause identified by the cause identifying device 6. Next, the operation of the abnormality diagnosis device in this embodiment will be described.

The process amount x of the plant 1 is input to the data collecting device 2 via the transmission line 8 and the data collecting device 2
At every sampling time interval Δt, a digital value (sample value) is converted and collected. Thereafter, the process quantity x converted into a digital value is transmitted to the cause candidate selection device 3 via the transmission line 9. Further, the process amount x is the transmission line 10
It is also transmitted to the recording device 5 via.

The cause candidate selection device 3 is a data collection device 2
When an abnormal change in the process amount x is detected at the time t1 shown in FIG. 2A from the sample value of the process amount x sent from the device, the causes (cause a, cause b) assumed as the cause of the abnormality are detected. , Cause c, ...)
The signals are sequentially transmitted to the inverse simulator 4 via the transmission line 11. At this time, the selection device 3 also transmits the process quantity x (= x1) at this time t1 to the inverse simulator 4 sequentially via the transmission line 11.

The recording device 5 is shown in FIG. 2A every time the sample value of the process amount x is sent from the data collecting device 2.
The values of the time t and the process amount x shown in are written and recorded in a storage device (not shown) such as a magnetic disk inside the device 5.

The inverse simulator 4 performs numerical calculation with the current time t1 at the start of abnormality diagnosis (abnormality detection time) and its process amount x1 (value of the actual waveform x (t) at time t1) as initial values. Run. That is, with respect to the time t, the reverse ascent tf (0 <tf <t1) from the current time t1.
Run the simulation over the range. In this simulation, a plant modeled by the mathematical formula (1) is applied. Further, when the simulation is actually executed, a mathematical model a (x, u, which is different for each cause, as shown in Formula (2), corresponding to the assumed type of abnormal cause (a, b, c, ...). ), B (x, u),
Apply c (x, u), .... This is because the assumed mechanism of the cause of the abnormality is mathematically modeled for each cause and built in the inverse simulator 4 in advance, so that the mathematical model applied when the simulation is executed in the simulator 4 also differs. Hereinafter, the procedure of the simulation executed in the opposite direction with respect to this time will be described with reference to the drawings. FIG. 3 is a flowchart of the simulation. Here, a case will be described as an example where the cause a is transmitted from the cause candidate selection device 3 as an assumed cause of abnormality.

First, the inverse simulator 4 sets the current time t1 transmitted from the cause candidate selecting device 3 and its process amount x1 as initial values (step S1). The simulator 4 selects a (x, u), which is a mathematical model corresponding to the cause a transmitted from the cause candidate selection device 3, as the function f of the mathematical formula (1). x, u) is set. dx / dt is calculated by equation (1) with t = t1 and x = x1 (step S
2). From the calculation result dx / dt and the following formula,
A time t before (past) by the time t and a process amount x at the time t are calculated (step S3). x = x− (dx / dt) Δt (3) t = t−Δt (4)

Next, it is determined whether the new time t obtained in step S3 is less than or equal to time tf (step S4). That is, it is determined whether the time t has reached the simulation completion time tf. Where t ≦ tf
If it is determined, the simulation for the cause a is completed, and for the next cause b, the simulation is executed with the initial values of the simulation t1 and x1 as in the case of the cause a. If t> tf is determined, t
The simulation is continued until ≤tf (step S2, step S3 and step S4).

At this time, the situation of the simulation is shown in FIG.
As shown in (a), different simulation waveforms xa (t) (xa waveform), xb (t) (xb waveform), xc (t) (xc waveform, corresponding to causes a, b, c, ... ), ... is obtained. This simulation waveform is sent to the cause identifying device 6 via the transmission line 12. At the same time, of the actual process amount recorded in the recording device 5,
Time interval tf ≤ t ≤ t1 during which the simulation was executed
The record x (t) (actual waveform) of the actual process amount in the same section as
It is sent from the device 5 to the cause identifying device 6 via the transmission line 13.

In the cause identifying device 6, simulation waveforms xa (t) (xa waveform), xb (t) (xb waveform),
xc (t) (xc waveform), ... And the actual waveform x (t) are compared, the matching waveform is identified, and the corresponding cause is identified as the cause of the generated abnormality. Whether the waveforms match or not is determined by the error area Si (i = a, b, c,
...) is evaluated by selecting the simulation waveform with the minimum. The cause of the abnormality is identified based on this evaluation.

[0025]

[Equation 2] The cause identification result of the cause identifying device 6 is sent to the display device 7 via the transmission line 14 and displayed.

[0026]

According to the present invention, the current time and its process amount are set as the initial values of the simulation, and then the simulation is executed in the reverse direction with respect to time for each of a plurality of possible abnormality cause candidates. By comparing the obtained simulation waveform and the actual waveform obtained from the recorded values of the actual process amount in the past, and by identifying the true cause candidate based on the comparison result,

Since the simulation can be executed with the current time and its process amount as the initial values, the initial values can be given accurately and the accuracy of the simulation itself can be kept good. Therefore, the simulation waveform of the process amount and the recording of the actual waveform are accurately compared, which has the effect of increasing the accuracy of identifying the cause of the abnormality.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an abnormality diagnosis device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a cause identifying method in the same embodiment in comparison with a conventional cause identifying method.

FIG. 3 is a flowchart for explaining a reverse simulation of a process amount in the same embodiment.

FIG. 4 is a flowchart of a simulation of a process amount performed in a forward direction with respect to conventional time.

[Explanation of symbols]

1 ... Plant, 2 ... Data collection device, 3 ... Cause candidate selection device, 4 ... Inverse simulator, 5 ... Recording device,
6 ... Cause identifying device, 7 ... Display device, 8-14 ...
Transmission line, xa to xc, xp to xr ... Simulation waveform.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Internal reference number FI Technical indication location // G06F 15/20 D 7052-5L (72) Inventor Takashi Hiramatsu Wadazaki, Hyogo-ku, Kobe-shi, Hyogo 1-1-1 Machi Mitsubishi Heavy Industries, Ltd. Kobe Shipyard

Claims (1)

[Claims] 1. An abnormality diagnosing device for diagnosing an abnormality in a process of various plants, wherein a current time and a process value are set as initial values of a simulation, and each of a plurality of possible abnormality cause candidates is timed. Regarding the above, the simulator that executes the simulation in the reverse direction is compared with the simulation value of the process amount, which is the execution result of the reverse simulation for each cause candidate of this simulator, and the past recorded value of the actual process amount. An abnormality diagnosis device comprising: a cause identification device that identifies a true cause candidate based on the above.