JPH0674874A - Abnormality generation time estimation device - Google Patents

Abstract

PURPOSE:To fully reduce an estimation error in estimating an abnormality generation time. CONSTITUTION:The amount of process inside a plant 1 is measured by a sensor 1-1, is converted to digital values and is collected by a data collector 2, and is transmitted to a diagnosis device 3, an inverse simulator 4, and a recorder 5. The cause of an abnormality is diagnosed by a diagnosis device 3 according to the fluctuation of the amount of process, a current time and the amount of process are set as initial values by the inverse simulator 4, and then the amount of process is simulated in inverse direction regarding time using a numeric expression model corresponding to the cause. A generation time estimation device 6 calculates the deviation between the simulation result and the recorded value of the amount of past process in the recorder 5 at each sampling time, determines the a time when the deviation exceeds a set limit value to be an abnormality generation time, and then displays it on a display 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an abnormality diagnosis system for various plants such as nuclear power, thermal power, and chemistry,
In particular, it relates to an abnormality occurrence time estimation device suitable for estimating the abnormality occurrence time of the plant.

[0002]

2. Description of the Related Art It can be said that the prior art has not obtained an effective means for estimating an abnormality occurrence time. As an example, in the case of a conventional system that detects an abnormality by issuing an alarm, a system as shown in FIG. 4 was typical.

In the example of the system shown in FIG. 4, the process quantity X = (x1, x2, x3, ..., Xn) of the plant 21 is sent to the data collecting device 22 via the transmission line 27 and converted into a digital value. Then, the alarm device 2 is transmitted via the transmission line 28.
Input to 3.

The alarm device 23 has a process quantity X = (x1,
When x2, x3, ..., Xn) exceeds the alarm set value Y = (y1, y2, y3, ..., Yn) set in the device 23, an alarm Z = (z1, z2, z3, ..., Zn). ) Is generated. That is, each variable xi of process amount X (i = 1, 1,
For 2, 3, ..., N), an alarm is issued by the following formula.

[0005]

[Equation 1] Here, when zi = 0 (i = 1, 2, 3, ..., N), no alarm is issued for the item i, and when zi = 1, an alarm is issued for the item i. .

Alarm information Z = (z1, z2, z3, ..., Z
n) is sent to the alarm recording device 24 via a transmission line 29,
Z of the alarm information Z = (z1, z2, z3, ..., Zn)
Only the item i of i = 1 is recorded in the storage device such as a magnetic disk in the recording device 24 together with the alarm issuance time ti. This alarm record (ti, zi) is sent to the analyzer 25 via the transmission line 30. The analyzer 25 analyzes the sent alarm record and searches for an alarm having the earliest alarm notification time ti (the alarm that is earliest in time). That is, each alarm issuance time t1 of the alarms z1 to zn
Of tn, the minimum value is set to ta, and the earliest alarm issuance time ta is regarded as the time immediately after the occurrence of the abnormality, and is estimated to be substantially equal to the abnormality occurrence time tf. The estimated abnormality occurrence time tf is sent to the display device 26 via the transmission line 31 and displayed. The above is a typical example of the abnormality occurrence time estimation system in the conventional technique.

[0007]

The above-described conventional abnormality occurrence time estimation system has a problem that the time required until the first alarm is issued is an estimation error of the abnormality occurrence time. That is, assuming that a certain process amount xi has a value of xio at the initial stage, when this process amount xi begins to change from the abnormality occurrence time tf and reaches and exceeds the alarm set value yi, an alarm i is issued at time ti. Then, the alarm information turns to zi = 1. Where ti (i = 1, 2, 3,
, N), the minimum value is set as the alarm time ta, as described above. At this time, the difference Δt between ta and tf occurs as an estimation error of the abnormality occurrence time. Δt = ta −tf (2)

The reason why such a problem occurs is that the difference between the normal value xio and the set value yi is usually large. This is because a sufficient margin is provided to prevent frequent false alarms. for that reason,
The time error Δt tends to be large.

The present invention has been made in view of the above circumstances, and an object thereof is to provide an abnormality occurrence time estimation device capable of sufficiently reducing an estimation error in estimating an abnormality occurrence time.

[0010]

According to the present invention, a simulator for simulating an abnormal state is set with a value of a current process amount as an initial value, and a simulation of the process amount is performed in a reverse direction with respect to time. Compare the simulated value of the process amount in the opposite direction with the recorded value of the actual process amount in the past, and if the deviation between the two is sufficiently large, the past time closest to the current time is defined as the error occurrence time. The feature is that it is estimated.

[0011]

According to the present invention, the simulation of the abnormal situation as shown in FIG. 2 is executed by the simulator in the reverse direction with respect to the time with the process amount value Xp of the current time tp as the initial value, and the result of the simulation is displayed. A simulation waveform Y (t) as indicated by a broken line in 2 is obtained. Further, in the figure, the recorded actual waveform X (t) of the actual process amount is indicated by a solid line. Waveform Y
(T) is the whole (0≤t≤tp) is a simulation waveform of an abnormal situation, whereas the actual waveform X (t) in which the actual process amount is recorded is from the abnormality occurrence time tf to the current time tp. Is abnormal (time region tf≤t≤tp) and normal in the time region (0≤t≤tf) before tf. Because of this difference, X (t) and Y (t) are different in the time domain (0≤t≤tf) before tf. Therefore, the difference between X (t) and Y (t) is the characteristic of the time domain (0≤t≤tf) before the occurrence of the abnormality.

The occurrence time estimation device uses X in the time domain (0≤t≤tf) before the occurrence of the abnormality, that is, in the normal domain.
The difference between (t) and Y (t) is the difference between the two, e (t) = |
Grasping with X (t) -Y (t) |, the time td at which e (t) starts to take a sufficiently large value is obtained.

As shown in FIG. 2, this time td is the time value that is positioned most chronologically behind in the normal region, and is therefore the closest to the abnormality occurrence time tf. Therefore, the occurrence time estimation device can estimate the abnormality occurrence time by approximating td to tf.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of the abnormality occurrence time estimation device according to the present embodiment. The sensor 1-1 is the plant 1
It measures the actual process amount of the data and is connected to the data collection device 2 via the transmission line 8.

The data collecting device 2 converts the actual process amount measured by the sensor 1-1 into a digital value and collects it. The data collecting device 2 is connected to the diagnostic device 3, the inverse simulator 4 and the recording device 5, respectively. 9, the transmission line 10 and the transmission line 11 are connected. The diagnostic device 3 is for diagnosing the cause of abnormality from the fluctuation of the process amount, and is connected to the inverse simulator 4 via the transmission line 12.

The inverse simulator 4 executes a simulation of an abnormal situation in a reverse direction with respect to time based on the cause of abnormality diagnosed by the diagnostic device 3.
3 is connected to the occurrence time estimation device 6. The recording device 5 records the value of the actual process amount transmitted from the data collection device 2, and is connected to the generation time estimation device 6 via the transmission line 14.

The generation time estimation device 6 is composed of the inverse simulator 4
The abnormality occurrence time is estimated by comparing the simulated value of the abnormal situation transmitted from the recording device 5 with the recorded value of the past actual process amount transmitted from the recording device 5, and the abnormality occurrence time is estimated through the transmission line 15. Connected to 7.
The display device 7 displays the abnormality occurrence time transmitted from the occurrence time estimation device 6. Next, the operation of the abnormality occurrence time estimation device having the above configuration will be described.

Process quantity in plant 1 X = (x1, x
The values of 2, x3, ..., Xn) are measured by the sensor 1-1.
The measured process amount X is transmitted to the data collection device 2 via the transmission line 8, converted into a digital value by the data collection device 2, collected, and provided for subsequent processing. The process quantity X digitally converted by the data collecting device 2 is transmitted to the diagnostic device 3, the inverse simulator 4 and the recording device 5 via the transmission line 9, the transmission line 10 and the transmission line 11, respectively.

In the diagnostic device 3, the process quantity X = (x1, x2, x3, ..., Xn transmitted from the data collection device 2).
) Diagnose the cause of abnormality from the aspect of change. The diagnostic device 3 diagnoses that, for example, if there is an abnormal variation in x1 of the process amount X, an abnormal cause a occurs, and if there is an abnormal variation in x2 of the process amount X, an abnormal cause b occurs. To do.
Appropriate existing methods can be applied to the diagnosis itself of such an abnormal cause. The cause of abnormality, which is the diagnosis result of the diagnosis device 3, is transmitted to the inverse simulator 4 via the transmission line 12.

In the inverse simulator 4, a mathematical model corresponding to the cause of the abnormality transmitted from the diagnostic device 3, for example, a differential equation of the time t with respect to the simulation waveform Y (t) represented by the following mathematical formula (3) is calculated in the time domain t. By performing a numerical operation in ≤tp, the simulation of the behavior of the actual process amount X (t) in the time domain t≤tp is executed. Here, tp is the current time, that is, the simulation execution start time. dY / dt = f (Y, u) (3) Here, f is a function relating to plant dynamic characteristics, and u is a disturbance input.

The simulation of the abnormal situation of the process quantity is performed in the reverse direction with respect to time. The calculation algorithm is as shown in FIG. As the initial value of the numerical operation performed at this time, the current values Xp and tp of the actual process amount input from the data collection device 2 via the transmission line 10 are set. The procedure of the simulation executed in the opposite direction with respect to this time will be described below with reference to the flowchart shown in FIG. Here, the diagnostic device 3
However, it is assumed that the cause of abnormality is diagnosed as cause a, for example.

The inverse simulator 4 selects, for example, a (Y, u) which is a mathematical model corresponding to this, based on the cause a transmitted from the diagnostic device 3, and as a function f of the mathematical formula (3). This a (Y, u) is set. The simulator 4 sets the process amount Xp at the current times tp and tp transmitted from the data collecting device 2 as an initial value of the simulation (step S1). next,
The inverse simulator 4 calculates dY / dt from the equation (3) with initial values t = tp and Y = Xp (step S).
2).

From the calculation result dY / dt and the following equations (4) and (5), the inverse simulator 4
The simulation value Y (=) of the process amount at the time t before (past) by the time t and the new time t.
Y (t)) is calculated (step S3). Y = Y- (dY / dt) Δt (4) t = t-Δt (5)

Next, the inverse simulator 4 determines whether or not the simulation has reached the time region (normal region) where the process amount is normal, depending on whether or not the estimated time device 6 has estimated the abnormality occurrence time. Yes (step S4).
Here, if it is determined that the normal region is reached (step S4), the simulation for the cause a is completed. If it is determined that the normal region has not been reached (step S4), the simulation is continued (steps S2, S3 and S4).

The time t which is the result of the above simulation
The simulation waveform Y (t) of the abnormal situation at (t≤tp) is calculated every time the inverse simulator 4 calculates,
It is transmitted to the occurrence time estimation device 6 via the transmission line 13.
On the other hand, the waveform X (t) of the actual process amount at time t is
The occurrence time estimation device 6 is read out via the transmission line 14 after reading what was written and stored as a recorded value of the actual process amount in the storage means such as the magnetic disk of the recording device 5.
Transmitted to. FIG. 2 shows an estimated situation in the occurrence time estimation device 6. In the figure, the vertical axis represents the process amount and the horizontal axis represents time.

In the occurrence time estimation device 6, each time the inverse simulator 4 performs a numerical calculation in the direction of decreasing the time from the current time tp, the time t sequentially transmitted from the simulator 4 is calculated.
And the simulation waveform Y (t) of the abnormal situation in
The recorded value X (t) of the actual process amount at the same time t transmitted from the recording device 5 is compared. The comparison is performed by focusing on the deviation value E according to the following equation. e (t) = | Y (t) -X (t) | ... (6) E = maximum value (| e1 |, | e2 |, ..., | en |) ... (7)

For the value of the simulation waveform Y (t) successively transmitted every time the inverse simulator 4 calculates in the direction of decreasing the time from the time point tp, the E value is investigated by the equations (6) and (7). The time td at which the E value exceeds a certain limit value F is recorded, and at the same time, the processing of the device 6 is terminated when the time td is reached. The occurrence time estimation device 6 estimates this time td as the abnormality occurrence time. Reverse simulator 4
Ends the simulation assuming that the normal region has been reached by estimating the abnormality occurrence time td of the occurrence time estimation device 6. The time td estimated as the abnormality occurrence time by the occurrence time estimation device 6 is transmitted to the display device 7 via the transmission line 15 and displayed on the display device 7.

[0028]

According to the present invention, the values of the current time and process amount are set as the initial values of the simulation, and the process amount simulation is performed in the reverse direction with respect to the cause of the diagnosed abnormality. Compare the resulting simulation value of the process amount with the past recorded value of the actual process amount, and set the current time at the time when the deviation between the simulation value and the recorded value of the actual process amount becomes larger than the preset limit value. By estimating the nearest past time as the time of anomaly occurrence,

The abnormality occurrence time can be accurately estimated, and the accuracy of the estimated time value of the abnormality occurrence time can be improved. Therefore, it is extremely effective in investigating the cause of occurrence of the cause of abnormality, and also has the effect of providing the operator with accurate information in plant operation and making the operation appropriate.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram of an estimation situation in the occurrence time estimation device 6 of FIG.

FIG. 3 is a flowchart of backward simulation with respect to time in the inverse simulator 4 of FIG.

FIG. 4 is a block diagram showing a configuration of a conventional abnormality occurrence time estimation system.

[Explanation of symbols]

1 ... Plant, 1-1 ... Sensor, 2 ...
Data collection device, 3 ... Diagnostic device, 4 ... Inverse simulator, 5 ... Recording device, 6 ... Occurrence time estimation device,
7 ... Display device, 8-15 ... Transmission line.

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁵ Identification number Office reference number FI technical display area // G06F 15/20 D 7052-5L

Claims (1)

[Claims] 1. In an abnormality diagnosis system for processes of various plants, a current time and a process amount are set as initial values of a simulation, and a process amount simulation is executed in a reverse direction with respect to a diagnosed cause. The simulator and the simulation value of the process amount obtained as a result of the simulation are compared with the past recorded value of the actual process amount, and the deviation between the simulated value and the recorded value of the actual process amount is a preset limit value. An occurrence time estimation device that estimates a past time that is the largest and is closest to the current time as an abnormality occurrence time, and an abnormality occurrence time estimation device.