JPH0682280A - Processing device for alarm - Google Patents

Abstract

PURPOSE:To remove the effect of a noise component and a drift component and to detect occurrence of abnormality more speedily and also reliably without outputting a false alarm. CONSTITUTION:This device is equipped with an input means 21 which reads an electric signal corresponding to a specific physical amount and subjects it to an input processing and with a movement average processing means 22 which executes a movement average processing for an input signal read by the input means 21. Moreover, it is equipped with a decision output means 24 which compares the signal subjected to the movement average processing by the movement average processing means 22 with an alarm set value set beforehand, decides an excess of a change of the signal over the alarm set value as abnormal and outputs an alarm outside.

Description

Detailed Description of the Invention

[0001]

The present invention relates to various plants,
An alarm processing device that monitors an electrical signal corresponding to a specific physical quantity (temperature, pressure, etc.) in the device, detects an abnormality in the state of the plant or device due to a change in the electrical signal, and outputs an alarm. In particular, the present invention relates to an alarm processing device that eliminates the influence of noise components and drift components and can detect abnormalities faster and more reliably without outputting false alarms.

[0002]

2. Description of the Related Art Generally, for example, in various plants and equipment, a specific physical quantity (temperature, pressure, etc.) is converted into an electric signal and monitored, and a change in the electric signal causes an abnormality in the state of the plant or equipment. An alarm processing device that detects the fact and outputs an alarm is often used. And
In this type of alarm processing device, the electric signal often contains a noise component and a drift component, and the following method has been conventionally adopted as a method for detecting the occurrence of an abnormality. (A) Method of comparing with raw data This method is a method of comparing an input electric signal with a predetermined alarm set value and generating an alarm when the input signal deviates from the alarm set value. However, this type of method has the following problems.

FIG. 6 is a diagram showing an example of an electrical signal containing a noise component and a drift component, and FIG. 7 is a diagram showing a state in which a signal change indicating an abnormal state of a plant or a device occurs in this electrical signal. That is, when the alarm set value is as large as ALH, the alarm is not output when the plant and the device are normal, and the alarm is output only in the abnormal state. However, also in this case, if there is a long-period signal fluctuation due to the drift component, an alarm may be output even when the degree of abnormality of the plant or the device is small (when the amount of change in the electric signal is small).

On the other hand, when the alarm set value is as small as ALL, an erroneous alarm may be output due to a noise component or a drift component even if the plant and equipment are normal. (B) Method of averaging

In this method, when the input signal contains a noise component of a relatively short period, the signal is averaged by a filter device to remove the noise component and then compared with the alarm set value.

In this case, as an averaging means, there are a case in which an averaging is performed by an electric circuit using a resistor and a capacitor, and a case in which an electric signal is read into a computer at a constant cycle and averaging is performed by a program process. There is. However, this type of method has the following problems.

FIG. 8 is a diagram showing a state in which noise components are removed by performing averaging processing on the electric signal of FIG. That is, in this case, the possibility of outputting a false alarm is less than in the case of the above (a), but when the alarm set value is small, there is still a possibility of false alarm output due to signal fluctuation due to the drift component. Remains. Further, when an abnormal state occurs, an alarm may be output even when the degree of abnormality is small (when the amount of change in the electric signal is small). (C) Method by averaging process + change rate determination This method is a method of performing abnormality determination by comparing with the previous average value after the averaging process by the method of (b). However, this type of method has the following problems.

FIG. 9 is a diagram showing this state. In addition,
For the sake of explanation, it is assumed that the reading cycle of the electric signal is t and 10
Averaging input values and averaging results A1, A2, A3
Shall get ... Also, when an abnormality occurs at time T,
The change of the electric signal due to the abnormality is set to 1V and the alarm set value is set to 1V.

In this case, the average calculation result A3 is increased by 0.7 V due to the influence of the abnormal signal. Therefore,
Even if the average calculation results A2 and A3 are compared, the abnormality cannot be detected. And in order to detect abnormalities,
It is necessary to compare the two previous results A2 and A4. In this case, the time from abnormality occurrence to detection (response time) is 1
It becomes 0t to 20t.

Further, in FIG. 9, it is assumed that the time (rise time) required for the change of the electric signal due to the abnormality is 0, but the rise time takes 2 t or more, and the abnormality occurs across the averaging calculation timing A3. Assuming that this is done, the abnormality cannot be detected even by comparing the average calculation results A2 and A4, and can be detected only by comparing the average calculation results A2 and A5. When the rise time is within 10t, the response time is 20t to 30t. Further, when the change amount of the electric signal due to the abnormality is very large, for example, when it is 2 V or more, the abnormality can be detected by comparing the average calculation results A2 and A3. Even if it is assumed that the change amount of the electric signal due to the abnormality is very large, the response time is 0 to 10 t. In this way, the abnormality determination timing may be delayed due to the averaging calculation, which may delay the abnormality response measures. (D) Method for measuring the cause of drift component

This method is the same as the above method (b),
This is a method of measuring an amount that causes a drift component and correcting the electric signal based on the measured value. That is, for example, when it is known that the cause of the drift component is temperature and the relationship between the temperature and the drift amount is known, a temperature detector can be attached and the electrical signal can be corrected by the temperature measurement value. This method can eliminate the drift phenomenon by correcting the electric signal,
It can be said to be an ideal method. However, this type of method has the following problems. That is, in one case, the cause of the drift component may not be clear, and in this case, the correction process cannot be performed.

On the other hand, even if the cause of the drift is clear, it may be difficult to measure the causative amount. Further, even if the measurement is possible, it is necessary to add a new measuring device.

[0013]

As described above, the conventional alarm processing device has a problem that an erroneous alarm is output or the detection of an abnormality is delayed.

An object of the present invention is to provide an extremely reliable alarm processing device which eliminates the influence of noise components and drift components and can detect the occurrence of an abnormality faster and more reliably without outputting a false alarm. To provide.

[0015]

In order to achieve the above object, according to the present invention, an input means for reading and inputting an electric signal corresponding to a specific physical quantity, and moving with respect to the input signal read by the input means. The moving average processing means for performing the averaging processing and the moving averaged signal by the moving average processing means are compared with a predetermined alarm set value, and an abnormality occurs when the signal change exceeds the alarm set value. And a determination output means for outputting an alarm to the outside.

Here, particularly as the moving average processing means, when the signal reading period is t and the number of readings is n, the past n times of average calculations are performed every time t, and the average calculation is performed every time t. I am trying to update the results.

As the judgment output means, the moving average calculation result by the moving averaging processing means is sequentially compared from the calculation result one time before to the calculation result a certain time before the rising time is assumed, and the alarm set value is set. An alarm is output when it exceeds.

[0018]

Therefore, in the alarm processing apparatus of the present invention, the moving average processing is performed on the electric signal corresponding to the specific physical quantity, the moving averaged signal is compared with the alarm set value, and When the amount of change exceeds the alarm set value, it is possible to eliminate the effects of noise components and drift components by determining that an abnormality has occurred and outputting an alarm. Can be detected more quickly and reliably.

[0019]

First, the concept of the present invention will be described.

As an averaging method for removing noise, there is a moving average method. This moving average method is a method in which, when the reading cycle is t and the number of readings is n, the average calculation is performed n times in the past every time t, and the average calculation result is updated every time t. The moving average method is generally used more frequently than the simple average method because the averaging result is smooth. Therefore, by using this moving average method, it is possible to detect the occurrence of an abnormality earlier by comparing the calculated value with a value calculated a predetermined time ago.

That is, FIG. 3 is a diagram showing a state of a moving average. Also in this case, the moving average of the values read 10 times in the cycle t is performed. As shown in FIG. 3, if an abnormality occurs, the moving average result changes stepwise as shown.

Therefore, the change of the electric signal due to the abnormality is 1
V and alarm set value 1V, to detect the alarm,
It is necessary to compare the result 10 times before. In this case, the response time is 10t. Furthermore, considering the case where the rise time is 5 t, it is sufficient to compare the calculated value 15 times before.
Furthermore, assuming a rise time of 10 t, 2
Compare it with the data 0 times before. In this case, the response time is 20t.

If the amount of change due to the abnormality is large, for example, if it is 5 V, the abnormality can be detected by comparing with the calculated value two times before. In this way, when the amount of change is large, the response time becomes short.

Therefore, in order to deal with various amounts of change, after the average calculation, it is sequentially compared from the calculation result one time before to the calculation result a certain time before the rise time is assumed, and the alarm set value is exceeded. If an alarm is output in the case of a response, the response time can be further shortened.
Comparing the response times of the simple average method and the moving average method is as follows.

Simple average method Moving average method Abnormal change amount = 1V, rise time = 10t 20t to 30t 20t Abnormal change amount = 10V, rise time = 0 0tt to 20t 1-2t

From the above points, in the present invention, the moving average processing is performed on the electric signal corresponding to the specific physical quantity, the moving averaged signal is compared with the alarm set value, and When the change exceeds the alarm set value, it is determined to be abnormal and an alarm is output. An embodiment of the present invention based on the above concept will be described below in detail with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of an alarm processing device according to the present invention.

That is, as shown in FIG. 1, the alarm processing apparatus of the present embodiment comprises an input means 21, a moving average processing means 22, a judgment output means, a judgment means 23 and an output means 24. ing.

Here, the input means 21 is a specific physical quantity (temperature, pressure, etc.) in various plants and devices not shown.
The electric signal corresponding to is read and input processing such as analog / digital conversion is performed. Further, the moving average processing means 22 performs moving average processing on the input signal read by the input means 21.

Further, the judging means 23 compares the signal averaged by the moving averaging processing means 22 with a predetermined alarm set value, and if the change in the signal exceeds the alarm set value, it is judged as abnormal. It is a judgment. Furthermore, the output means 24 comprises, for example, a contact output device,
The output of the abnormality determination from the determination means 23 is output to the outside as an alarm. Next, the operation of the alarm processing device of the present embodiment configured as described above will be described using the flow chart shown in FIG.

In FIG. 1, the input means 21 receives electric signals corresponding to specific physical quantities in various plants and devices, that is, output signals from various sensors,
Here, analog / digital conversion processing is performed and the result is input to the moving average processing means 22. Next, moving average processing means 2
In 2, the moving average calculation of the signal input from the input means 21 is performed.

Next, in the judging means 23, the signal averaged by the moving average processing means 22 is compared with a predetermined alarm set value. If the change in the signal exceeds the alarm set value, it is determined to be abnormal, and the abnormality determination output is made. Then, the output means 24 outputs the abnormality determination output from the determination means 23 to the outside as an alarm.

In this case, the moving average processing means 22
The moving average calculation of the signal is performed as follows, for example. In addition, here, in order to simplify the description, the following example will be described.

Sampling period (t) = 6 seconds Number of averaging data (n) = 10 (that is, averaging period = 1
Min) Signal increase due to leakage = 1V Alarm set value = 1V

FIG. 3 is a diagram showing an outline of the moving average method. That is, in the present embodiment, as shown in FIG. 3, the signal is sampled every 6 seconds, and the data of the past 10 times are averaged. Therefore, the average calculation result is updated every 6 seconds.

Generally, the moving average method is used more often than the simple average method because the averaging result is smooth. Figure 3
In, it is assumed that a leak (abnormality due to signal fluctuation is expressed as a leak) occurs between cycles 5 and 6. In this case, the averaging result changes stepwise as shown in FIG. Therefore, the leakage cannot be detected by comparing the adjacent data. So to detect a leak,
It is necessary to compare with 10 times before (data 5 and 15). In this case, the response time is 1 minute. Furthermore, consider the case where the rise time is not 0 seconds. If the rise time is 30 seconds, it is sufficient to compare five times before (data 5 and 20). If the rise time is assumed to be 1 minute, the data 5 and 25 may be compared. In this case, the response time is 2 minutes from the start of leakage.

On the contrary, considering that there is a large amount of leakage instantaneously, after calculating the data 25, it is compared with the data from 24 to 5 in order, and if the difference is 1 V or more, an alarm is output. Is the best way to do it.

In this way, a signal is input every 6 seconds,
The moving average method is performed in which the input signals of the past 10 times are averaged and the averaging calculation result is calculated, that is, the averaging result is updated every 6 seconds. Then, the calculation result every 6 seconds, the calculation result of one time before, the calculation result of the second time before ... The calculation result of 20 times before are sequentially compared, and if the calculation result of this time is increased by 1 V or more, it is determined as abnormal. To be done. In this case, when the rise time of the abnormality is within 1 minute, the abnormality is caught within the response time of within 2 minutes. In addition, when the amount of change due to an abnormality is large (3V to 5V, etc.), the response time is improved according to the change of 24 to 10 seconds. Next, the fact that the moving average method of this embodiment is better than the simple average method will be described. The simple average method will be described with reference to the flowchart shown in FIG.

As shown in FIG. 5 (a), when the latest averaging result, for example B, is calculated, it is compared with the immediately preceding calculation result A, and if there is a change over the set value, leakage has occurred. To determine.

As shown in FIG. 5 (b), it is assumed that there is a leak in the middle of a, there is a signal increase of 1V, and the rise time is 0 seconds. The averaging result at this time is A = 0, B = 0.4, and C = 0.6, as shown in the figure. In this case, even if A and B are compared and B and C are compared, leakage cannot be detected as long as the alarm set value is 1V. Therefore, in order to detect the leakage, it is necessary to set the alarm set value to 0.5V or to compare A and C, that is, the average value two times before.

When the alarm set value is 0.5 V, the response time is 1 minute 30 seconds or less, but the alarm is issued even though the amount of leakage is small (leakage of 0.5 V) depending on the timing of leakage occurrence. It is possible that it will occur. Also,
When the alarm set value is set to 1 V, it is necessary to compare two times before (A and C) to detect leakage. In this case, the maximum response time is 2 minutes. Considering that the response time is not a problem when comparing the two, the latter (comparison with two times before) is superior in order to avoid unnecessary alarms. On the other hand, the rise time of the alarm is not 0 seconds,
The case of 6 seconds or more is assumed. Now, in FIG. 5B, 9 = 0.25V for a, 10 = 0.5V for a, 1 = 0. For C.
Assume that 75V and 2 = 1V of C. In this case, the average result is: A = 0 B = 0.075 C = 9.975 D =
Therefore, as long as the alarm set value is 1 V, even if A and C are compared and B and D are compared, leakage cannot be detected.

In order to detect leakage, it is necessary to compare three times (A and D). In this case, the response time is 3
It will be a minute. If this method is adopted, the rise time is 1
If it is less than a minute, it cannot be detected unless the response time is 3 minutes. From the above viewpoints, the response times of the moving average method and the simple average method are compared as follows.

Simple averaging method Moving average method Leakage amount = 1 V, rise time = 1 min 2-3 min 2 min Leakage amount = 10 V, rise time = 0 min 1-2 min 6-12 sec Thus, the moving average method It is clear that the response is faster and the response time has less variation.

As described above, the alarm processing apparatus according to the present embodiment has various physical quantities (temperature,
Input means 21 for reading an electric signal corresponding to pressure and the like and performing input processing such as analog / digital conversion, and a signal reading cycle of the input signal read by the input means 21 is t and the number of readings is n. In such a case, the moving average processing unit 22 and the moving average processing unit 22 perform the moving average processing for performing the average calculation of the past n times at each time t and updating the average calculation result at each time t. The moving average calculation result is sequentially compared from the calculation result one time before to the calculation result a certain time before the rise time is assumed, and it is judged as abnormal when the alarm set value is exceeded 23
And an output unit 24 that outputs the abnormality determination output from the determination unit 23 to the outside as an alarm.

Therefore, since it is possible to eliminate the influence of the noise component and the drift component, it is possible to generate an abnormality more quickly without outputting an erroneous alarm depending on the size of the alarm set value as in the conventional case. Since it is possible to reliably detect the alarm processing device, the alarm processing device can be provided with extremely high reliability.

In the above embodiment, the moving average processing means 22 performs the past n times of average calculation every time t when the signal read cycle is t and the number of read times is n. Although the case of updating the average calculation result has been described above, it goes without saying that the present invention is not limited to this.

Further, in the above embodiment, the judgment and output means 23 and 24 are the calculation results of the moving average calculated by the moving averaging processing means 22 from the calculation result of one time before and the calculation result of a certain time before the rise time is assumed. The case where the alarm is output when the alarm set value is exceeded has been described above, but it goes without saying that the present invention is not limited to this.

[0047]

As described above, according to the present invention, the input means for reading and inputting an electric signal corresponding to a specific physical quantity, and the moving average processing for the input signal read by the input means. The moving averaging processing means and the signal averaged by the moving averaging processing means are compared with a preset alarm set value, and it is determined that an abnormality occurs when the change in the signal exceeds the alarm set value. Since it is configured with a judgment output means for outputting an alarm to, it is possible to eliminate the influence of noise components and drift components, and to detect abnormalities faster and more reliably without outputting false alarms. A highly reliable alarm processing device can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an alarm processing device according to the present invention.

FIG. 2 is a flow chart for explaining the operation in the same embodiment.

FIG. 3 is a schematic diagram for explaining a moving average method.

FIG. 4 is a schematic diagram for explaining a simple averaging method.

FIG. 5 is a schematic diagram for explaining a simple averaging method.

FIG. 6 is a diagram showing an example of an electric signal including a noise component and a drift component.

FIG. 7 is a diagram showing a state in which a signal change indicating an abnormal state of a plant or a device occurs in the electric signal of FIG.

8 is a diagram showing a state in which a noise component is removed by performing averaging processing on the electric signal of FIG.

FIG. 9 is a diagram showing a state in which abnormality determination is performed by comparing with a previous average value after averaging processing.

[Explanation of Codes] 21 ... Input Means, 22 ... Moving Average Processing Means, 23 ... Judgment Means, 24 ... Output Means.

Claims (3)

[Claims] 1. Input means for reading and inputting an electrical signal corresponding to a specific physical quantity, moving averaging processing means for performing moving averaging processing on the input signal read by the input means, and the moving average. Determination output means for comparing the moving averaged signal by the conversion processing means with a predetermined alarm set value, and determining that the signal is abnormal due to a change in the signal exceeding the alarm set value, and outputting an alarm to the outside. An alarm processing device comprising: 2. The moving averaging processing means, when the signal reading period is t and the number of readings is n, performs an average calculation of past n times at each time t, and an average calculation result at each time t. The alarm processing device according to claim 1, wherein the alarm processing device is updated. 3. The judgment output means sequentially compares the moving average calculation result by the moving averaging processing means from the calculation result one time before to the calculation result a certain time before the rising time is assumed, and sets the alarm set value. The alarm processing device according to claim 1, wherein an alarm is output when the value exceeds.