JP2002032243A - Monitoring value processing method - Google Patents

Abstract

(57) [Summary] An object of the present invention is to avoid unnecessary trouble handling operation caused by noise generated in a monitored value. A process for periodically acquiring information from a monitoring target, a process for comparing a predetermined value with the information,
A process of determining and confirming the validity of the result of the comparison process,
In the monitoring process for causing an operation according to the confirmation process,
The determination process uses a majority decision process for a total of three pieces of information, the information obtained for the latest determination and the information obtained in the past two times in succession.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a monitoring value processing method in a computer system.

[0002]

2. Description of the Related Art In general, a monitoring operation means that the state of a monitored object 81 such as temperature and voltage is measured by a sensor 8 as shown in FIG.
2 and continuously converted into a fetched signal, and the signal is read by the monitoring processing system 83 on the detection side at regular intervals. In the monitoring processing system 83, a threshold value for distinguishing between normal and abnormal signal values is prepared in advance. When a signal from the monitored object is read, the magnitude relationship between the signal value and the threshold value is compared, and the normal value is determined. Alternatively, an abnormal value is determined. If it is determined to be abnormal, a failure handling operation 84 such as failure notification, failure recording, and system termination is induced.

If the result of the normal / abnormal judgment of the read signal is used as it is for inducing a failure response operation, the failure response operation may be induced even if an abnormal value due to noise unrelated to the existence of the failure occurs. And contribute to the instability of the system.

[0004]

Conventionally, in the process of determining the result of normality / abnormality determination, the previous value and the current value of the result of normality / abnormality determination are compared as a determination process. There has been a method in which the determination result is set as a fixed value, and when both are different values, an intermediate value between the two is set as a fixed value.

FIG. 4 shows an example of this processing. Reference numeral 61 denotes the result of the normal / abnormal determination described above. In this example, while the normal values are continuously determined, a value other than the normal value appears only in one sample in sampling # 3.

[0006] Such a pattern that differs from a series of values before and after only one sample is interpreted that the one sample is not a value detected due to a failure to be monitored, but a noise. That is, if the monitored value fluctuates from a normal value to an abnormal value due to a fault, the sampling period is set finer than the time required for recovery from the occurrence of the fault, and the abnormal value is set so that the change is reliably captured. Try to detect multiple times. Therefore, in the sampling period, the appearance time of only one sample as in this example is extremely short, and therefore, it may be determined that the sample is noise. Therefore, the value of this sampling # 3 is a value due to noise and does not indicate a failure to be monitored, and is a value to be processed as a normal value by the following determination processing of 62.

The description of the symbols used in 62 is shown in FIG. The two values located above the symbol are candidates for judgment,
The value located below the symbol indicates the determination result. If the previous value and the current value are equal, the value is determined as a determined value, and if different, the intermediate value thereof is determined as the determined value.

Reference numeral 63 denotes the result of the determination processing of 62. Even with the above processing, the level of the singular value of sampling # 3 approaches the normal value but does not return to the normal value, and this conventional technique is an incomplete process.

[0009]

SUMMARY OF THE INVENTION In order to improve the above-described conventional determination process, the present invention provides a process of periodically acquiring information from a monitoring target, and a process of obtaining information from a predetermined value and the information. In the comparing process, the process of determining and determining the validity of the result of the comparing process, and the monitoring process for performing an operation according to the determining process, in the determining process, the information used for the determination is obtained latest. The majority processing is performed on a total of three pieces of information, that is, information obtained and information obtained two times in a row in the past. This processing ignores a fluctuation value that exists only for one period and is regarded as noise in a monitoring value having a constant value, but does not affect sampling of a value generated for a plurality of periods.

The monitoring processing of the present invention has an effect of preventing an unnecessary trouble handling operation caused by noise from occurring.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment of the present invention will be described in detail with reference to the drawings.

In the monitoring operation, as shown in FIG. 6, a reference value 91 with an allowable range 92 is set for a monitored object, and when a monitored value at a certain time is included in this allowable range, 93 is detected as a normal value. , The operation of detecting the deviation 94 as an abnormal value. The abnormal value also serves as a trigger for a failure response operation such as a failure notification, a failure record, and a system termination.

As shown in FIG. 7, the change in the monitored value
It is characterized by the deviation amount L from the reference value and the deviation time τ. Now, a change in which both or one of these L and τ is minute is interpreted as noise that is not a change caused by a failure to be monitored, and must be ignored. That is,
When a failure actually occurs, the sampling period is set so that the abnormal value is sampled a plurality of times during the period of occurrence. It is assumed that a small change in τ that is sampled only zero or one time during that period does not cause a failure even if the deviation L from the reference value is an abnormal value.

Hereinafter, a method of nullifying such fluctuation interpreted as noise will be described.

First, a minute change in the deviation L from the reference value can be ignored by setting an allowable range of the normal value. On the other hand, the minute fluctuation of the fluctuation time τ can be ignored by setting the sampling period t0 of the monitoring value longer than the fluctuation time τ considered as noise as symbolically shown in FIG. As shown, the sampling timing is not avoided depending on the position of the fluctuation start time τ0, and sampling may occur for one timing.

The present invention shows a process for ignoring a singular value that occurs for one timing in a certain series of steady monitoring values, and at the same time, has an effect on sampling of a value that occurs a plurality of times in the same process. A process not to be given is performed.

1 shown in FIG. 1 indicates a time change of the monitored information such as a voltage value converted into an electric signal by a voltage sensor or the like. Hereinafter, this signal is referred to as a monitored value. Now, this pattern has two variable patterns 2
1 and 31, and the fluctuation time of the fluctuation pattern 21 is τ
1. The variation time of the variation pattern 31 is τ2. τ1 is a minute time, for example, 1 second. τ2 is a time sufficient for the observer to recognize the change, for example, one hour.

Next, the monitored value of the sensor output is read into the monitoring processing system. Reading (sampling) is performed discretely at a constant cycle. Now, with respect to the fluctuation time, the monitor specifies that the fluctuation of the monitored value within the fluctuation time of 2 seconds or less does not indicate the abnormality of the monitoring target, but is defined as noise. Therefore, the fluctuation 21 having a fluctuation time of 2 seconds is noise according to the definition of the observer. Fluctuation time 31 long enough for fluctuation time 31
Is a normal fluctuation value. The sampling period 10 is
To avoid noise detection, the fluctuation time is set to a time longer than 2 seconds, for example, a 4 second cycle. According to the above setting, the monitoring processing system samples the monitored value.

As a result of sampling, fluctuation 2 which is noise
1 is detected as a sample value 22 that has fluctuated for one cycle because the fluctuation start time τ0 is close to the sampling timing despite the fact that the fluctuation time τ1 is very small.

Next, in the monitoring processing system, the monitored value read in accordance with the above is compared with a predetermined threshold value. Now three regions by two thresholds 51, 52
(a), (b), and (c) are defined. In general, the presence of n thresholds defines n + 1 regions. The monitored value is compared with a threshold value, and the monitored value is converted into a region value corresponding to the region to which the monitored device belongs, thereby simplifying the value. The monitored value at the end of this processing has a fluctuation value 24 due to noise. If the monitored value at this point is used as a trigger for a failure response operation, an unnecessary failure response operation by the 24 Is induced.

Therefore, a determination process is further performed on the monitored value. In the determination process, a majority decision is performed on three monitored values in combination with the monitored values sampled at a certain point in time and the two monitored values that are continuous with the monitored values, and the result is determined as a determined value. As shown in FIG. 2, the symbol shown at 41 or the like has three values located at the top of the symbol, Value 1, Value 2, and Value 3.
A value of 4 which is a judgment candidate and is located below the symbol indicates the result of majority judgment.

Symbols 41 to 43 denote determination processing for the 24 fluctuation values caused by noise. As clearly shown in the pattern 60 after the finalization processing, the fluctuation value that exists only for one cycle in a series of monitoring values that take a constant value is determined by the majority decision processing to be the same as the constant value before and after the fluctuation. Patterns that have been equalized while maintaining a variable value over multiple periods are preserved.

[0023]

As described above, the monitoring value processing method realized by the present invention makes it possible to avoid unnecessary trouble handling operations caused by noise generated in monitoring values.
This is effective in preventing malfunctions in failure processing such as system stoppage or issuance of a warning bell, and can improve monitoring reliability. In addition, by avoiding unnecessary recording for the recording of a failure, which is one of the failure handling operations, the memory area allocated for the recording can be effectively used. This is particularly useful for long-term monitoring operations that last several years.

[Brief description of the drawings]

FIG. 1 is a diagram showing the contents of the invention of claim 1;

FIG. 2 is an explanatory diagram of symbols used in FIG.

FIG. 3 is a flowchart of a series of processes in a monitoring process.

FIG. 4 is an explanatory diagram of a conventional technique.

FIG. 5 is an explanatory diagram of symbols used in FIG.

FIG. 6 is a diagram showing a target of a monitoring operation.

FIG. 7 is a model diagram of monitored values.

FIG. 8 is a diagram showing an example of sampling a monitored value that fluctuates for a very short time.

FIG. 9 is a diagram showing an example of sampling a monitored value that fluctuates for a very short time.

[Explanation of symbols]

10: sampling period in the example of the monitoring process of the present invention, 21: monitored value that fluctuates minutely, 22: sampling of the monitored value that fluctuates minutely, 23: sampling value of the monitored value that fluctuates minutely, 24 ... Sampling values of monitored values that fluctuate minutely, 41: majority processing for monitored values that fluctuate for a small time, 42: majority processing for monitored values that fluctuate for a small time, 43: majority processing for monitored values that fluctuate for a small time, Reference numeral 60: a part where the fluctuation has been eliminated by the monitoring value processing method of the present invention; 61, a pattern diagram of monitored values immediately before determination processing in the example of the conventional technique;
... Determination processing in the example of the conventional technology, 63... Pattern diagram of the determination processing result in the example of the conventional technology, 81... Environmental factors to be monitored in the monitoring processing flow, 82. 84, a failure handling operation in the monitoring processing flow, 91, a reference value of the monitored value, 92, an allowable range associated with the reference value of the monitored value, 93, a sampling that is regarded as a normal value with the monitored value. Points, 94: sampling points at which monitored values are regarded as abnormal values, 98: sampling examples of monitored values that fluctuate for a short time.

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Daiki Abe 1 Horiyamashita, Hadano-shi, Kanagawa F-term (reference) 5B034 AA05 DD02 5B042 GC16 JJ08 JJ17 KK01 KK13 KK14 LA16 LA19 MC19

Claims (1)

[Claims] A process for periodically acquiring information from a monitoring target; a process for comparing a predetermined value with the information; a process for determining the validity of a result of the comparison process and determining the result; In the monitoring process for performing an operation in accordance with the determination process, the determination process includes a majority decision process for a total of three pieces of information, the latest information obtained for the information used for the determination and the information obtained two times in a row in the past. A monitoring value processing method characterized by being used.