JPS5916289B2 - Abnormal diagnosis method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an abnormality diagnosis method for definitively detecting an abnormality in a process, a measurement control device, or the process itself in various process controls.

Recently, there has been a strong demand for abnormality diagnosis of various types of process measurement and control equipment, etc., for the purpose of safe process operation and efficient maintenance and inspection. In particular, if we can detect an abnormality in fluoresce, a failure in measurement and control equipment, or an abnormality before it becomes definite, that is, when it has become slightly abnormal or is in the process of becoming a failure, it will be easier to take countermeasures.
This is also essential for safe process operation. Conventionally, process abnormality detection means have been used to detect process abnormalities by checking the upper and lower limits of process variables, and checking the validity of the detector...for example, setting the measurement range to 1 to 5 VDC. and when the signal is outside the DC1~5V range, that is, DC
It is relatively easy to detect the failure of a detector that has performed OV or 6V DC (the detector is judged to be abnormal), but in this case, the result of the failure is being detected, so post-processing is required. , process abnormalities and failures become certain.

However, the actual requirement is that the process itself or the process measurement and control equipment is becoming slightly abnormal, or that the process is being detected at the stage of a progressive abnormality that is gradually progressing, that is, before it becomes a definite failure. The key is to maintain safety and take proactive measures before the process goes down. However, in practice, it is extremely difficult to detect small or progressive abnormalities.

In view of the above points, the present invention provides an abnormality diagnosis method that reliably detects abnormalities and failures in processes and the like at a stage when these abnormalities and failures have definitely occurred.

That is, the present invention utilizes information from two or more locations in a process to compare a certain process variable with an estimated calculated value obtained by estimating the process variable from one or more other process variables that have a correlation with the process variable. , performs statistical processing based on the difference between the variables and estimated calculated values, reliably detects slight abnormalities, progressive abnormalities, etc., thereby ensuring safe process operation and preventing deterioration of operation.

Note that in order to accurately detect small abnormalities, progressive abnormalities, etc. as early as possible, the signal processing means for detection becomes a problem.

Naturally, when comparing process variables and their estimated calculated values, we correct for dead time and time constants that exist between them.
Although transient and theoretically consistent calculations are performed, actual process values do not completely match estimated calculated values based on correlated process values due to sudden changes and indirect effects of other conditions. Even so, simply checking the tolerance limit based on the difference between the two values will result in many false positives. Therefore, in order to solve these problems, it is necessary to perform statistical processing of the signal from the viewpoint of removing the influence of signal changes such as disturbances and noise, and properly understanding the slow progress abnormality.

Embodiments to which the present invention is applied will be described below with reference to the drawings.

In this embodiment, as statistical processing of signals, an example will be given in which set ratios and actually measured ratios of process pressure balance, flow rate balance, and energy balance are used. First, FIG. 1 shows a first embodiment to which the present invention is applied, in which 1 is a fluid supply pipe arranged on the tank input side, and the fluid pressure in this pipe 1 is detected by a pressure detector 2. .

The output end of the fluid supply pipe 1 is connected to the tank 3, thereby supplying fluid into the tank 3. Reference numeral 4 denotes a fluid delivery pipe disposed on the tank output side, which theoretically has a piping resistance 5, and also has a flow rate detector 6 sandwiched in its middle for transmitting a signal F proportional to the fluid flow rate. .

Furthermore, a pressure detector 7 is connected to the fluid delivery pipe 4 to detect the fluid pressure within the pipe 4 . Reference numeral 8 denotes an adder that adds the signal P1 proportional to the pressure from the pressure detector 7 and the pressure loss KF2 from the pressure loss calculator 9 connected to the flow rate detector 6.

The pressure loss calculator 9 calculates the pressure loss due to the flow rate from the pressure detector 2 to the pressure detector 7. The addition signal (P1+KF2) of this adder 8 is supplied to the subsequent response compensator 10. This response compensator 10 compensates for dead time L, time constant T, etc. 11 is a subtracter that obtains the signal difference between the pressure detector 2 and the response compensator 10, and the difference signal (PO
-Pc) is supplied to the comparator 12.

This comparator 12 is supplied with a predetermined tolerance value δ from the difference tolerance value setter 13 in advance, and therefore the difference signal (PO-
Pc) and the difference tolerance δ are compared. Further, two time integrators 14 and 15 are connected to the output side of the comparator 12, and one time integrator 14 has the output of the comparator 12 (PO-Pc
)>δ, and the other time integrator 15 integrates the time during which the output of the comparator 12 becomes (PO−Pc)<−δ. 16 is an integrated time difference T between both time integrators 14 and 15;

18 is the integrated time difference T. and the cumulative time tolerance T. It is a comparator that compares the 19 is the cumulative time allowable value T.

20 is the time integrator 1.
This is a sample period oscillator that resets 4 and 15 at regular intervals. Next, the operation of the apparatus shown in FIG. 1 will be explained.

Fluid is circulated or sent to a predetermined location by pipes 1 and 4 disposed within the process;
The pressure of the fluid inside the pipe is detected by pressure detectors 2 and 7 provided at predetermined locations in the pipe (which may be one pipe), and a signal P is proportional to the detected pressure. , Pl are output. Then, the output signal PO of the pressure detector 2 is sent to the subtracter 11, and the output signal PO of the pressure detector 7
The output signals P1 are respectively supplied to adders 8. In addition, a flow rate detector 6 provided on the fluid delivery pipe 4 on the tank output side
A signal F proportional to the fluid flow rate is output. Then, the signal F from the flow rate detector 6 is used to calculate the pressure loss KF2 for the flow rate by the pressure loss calculator 9, which is added to the signal P1 from the pressure detector 7 by the adder 8, and the pressure loss is calculated by the pressure loss calculator 9. Calculated pressure signal (P1+KF
2) is obtained. Furthermore, this pressure signal (P1+KF2) is compensated for dead time L, time constant T, etc. by response compensator 10, and the pressure P, which is a process variable, is determined. Estimated calculated value P
c is obtained, which is supplied to the subtractor 11. Therefore,
This subtractor 11 is the pressure P. A subtraction operation is performed between and the estimated calculated value Pc, and the pressure difference (PO-Pc) is taken out and guided to the subsequent comparator 12. On the other hand, this comparator 12 is given a predetermined difference tolerance value δ from the difference tolerance value setter 13 in advance, and therefore, the difference tolerance value δ and the pressure difference (PO-Pc) of the subtractor 11 are calculated in advance. A comparison is made.

As a result of this comparison, if (PO-Pc)>δ, the time integrator 14 integrates the time to maintain that condition, and (PO-Pc)
When <-δ, the time integrator 15 integrates the time during which the condition is maintained, and outputs the respective time integrated values Tp and tN.
Note that the unit time for time integration is obtained by resetting the time integrators 14 and 15 at predetermined time intervals using the sample period oscillator 20. Once the time integrated values TP and tN are obtained by the time integrator 14 and 15 in this manner, they are input into the subtracter 16 to obtain the integrated time difference T.

After determining =(Tp-TN), it is supplied to the comparator 18.
Then, the integrated time difference tolerance T is supplied in advance by this comparator 18. and as a result of this comparison, TO≧
In the case of TO, an abnormality signal A is generated for the process and the like. In addition, in the description of the above embodiment, the difference between the time integrated value when exceeding the positive side and the time integrated value when exceeding the negative side is detected, but for example, as these means,
Similarly, check which side the pulses are exceeding at regular intervals, find out the difference between the number of pulses exceeding the positive side and the number of pulses exceeding the negative side, and detect an abnormality when this difference exceeds a predetermined limit. A signal may also be transmitted.

Next, FIG. 2 shows a device to which another embodiment of the present invention is applied, which is a partially improved version of the device shown in FIG.

That is, in this embodiment, the pressure P extracted by the subtractor 11. and the estimated calculated value Pc (PO-Pc) is calculated by the integrator 2.
1, and the comparator 24 compares the integral value with the integral tolerance value of the integral tolerance value setter value 23, and generates an abnormal signal when the integral value exceeds the integral tolerance limit. be. 22 is a sample period generator.

Further, a process to which the present invention is applied may be carried out as shown in FIG.

That is, for each cycle transmitted by the signal cycle oscillator 31,
The pressure difference (PO-Pc) from the subtractor 11 is checked by the difference check circuit 32, and the distribution pattern of the difference and frequency within a certain period of time is determined and stored in the difference-frequency distribution pattern storage 33. A frequency distribution pattern comparator 34 compares changes in the patterns, and when the changes in the patterns exceed a predetermined allowable limit, an abnormality signal is generated.
In particular, in this embodiment, it is possible to accurately detect an abnormality in the extremely slow progress. It should be noted that the present invention is not limited to the above-mentioned embodiment, and therefore, although the above-mentioned analog calculation was explained, it may be performed by, for example, a time-division digital calculation by a computer.

It goes without saying that various other modifications can be made without departing from the spirit of the invention. As detailed above, according to the present invention, the difference between the process variable at a certain location and the estimated calculated value of the process variable from the perspective of the process variables at other locations that have a correlation therewith is calculated. It is possible to compensate for losses associated with flow rate and other types at the stage of this estimated calculated value, and furthermore, the difference between these process variables and the estimated calculated value can be calculated using the time difference within a fixed period (
Statistical processing is performed to diagnose the presence or absence of an abnormality by comparing it with a predetermined tolerance value as an integral value (positive, negative), etc., so even a slight abnormality or a progressive abnormality can be quickly and accurately detected. Therefore, it is possible to ensure the safe operation of the process, and by taking measures in advance in the event of an abnormality, it is possible to minimize the decline in operation.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing one embodiment of a process-based abnormality diagnosis device to which the present invention is applied, and FIGS. 2 and 3 are partial block diagrams of devices to which other embodiments of the present invention are applied. It is a diagram. 2...Pressure detector, 6...Flow rate detector, 7...Pressure detector, 8...Adder,
9...Pressure loss calculator, 10...Response compensator, 11...Subtractor, 12...Comparator, 13... Difference tolerance value setter, 14, 15・
...Time integrator, 18...Comparator, 19
・・・・・・Tolerance value setter for integrated time, 20・・・・・・
Sambra periodic oscillator, 21...Integrator, 22.
...Gumbra periodic oscillator, 23... Integral tolerance value setter, 24... Comparator, 31...
...Gumbra periodic oscillator, 32...Difference check circuit, 33...Difference frequency distribution pattern memory,
34... Comparator.

Claims (1)

[Claims] 1 A process variable detection means for detecting a process variable at a certain point on the process line and a process variable loss at one or more other places having a correlation with the process variable on the process line, A process variable loss calculation means for obtaining a calculated process variable at each point, and an estimated calculated value of the process variable by compensating for dead time, time constant, etc. for the calculated process variable obtained by this means. response compensation hand
The difference between the process variable detected by the process variable detection means and the estimated calculated value obtained by the response compensation means is determined, and this difference is calculated as a time difference, an integral value, or a frequency distribution pattern within a fixed period. etc., and compares this with a predetermined tolerance value, and if the time difference, etc. exceeds the predetermined tolerance value, the process or process measurement control equipment is diagnosed as being abnormal. Characteristic abnormality diagnosis method.