JPH07114483A - Fault diagnosing device - Google Patents
Fault diagnosing deviceInfo
- Publication number
- JPH07114483A JPH07114483A JP5257971A JP25797193A JPH07114483A JP H07114483 A JPH07114483 A JP H07114483A JP 5257971 A JP5257971 A JP 5257971A JP 25797193 A JP25797193 A JP 25797193A JP H07114483 A JPH07114483 A JP H07114483A
- Authority
- JP
- Japan
- Prior art keywords
- alarm
- failure
- certainty factor
- processing unit
- certainty
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Test And Diagnosis Of Digital Computers (AREA)
- Monitoring And Testing Of Exchanges (AREA)
Abstract
Description
【0001】[0001]
【産業上の利用分野】本発明は、故障診断対象に設置さ
れた警報装置から各種警報が発生した場合に、その警報
を分析することにより、前記故障診断対象を構成する各
装置の中で、どの装置が故障の可能性が高く、どの装置
が低いか、を判定する故障診断装置に関するものであ
る。BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention, when various alarms are generated from an alarm device installed in a failure diagnosis target, analyzes the alarms to thereby detect, among the devices constituting the failure diagnosis target, The present invention relates to a failure diagnosis device that determines which device has a high possibility of failure and which device has a low probability.
【0002】[0002]
【従来の技術】従来、故障診断対象に設置された警報装
置から各種警報が発生した場合に、その故障診断対象を
構成する各装置の中で、どの装置が故障の可能性の高い
被疑装置であるかを診断する方式として、発生した複数
の警報間の関連に着目して、その中から警報発生パター
ンを分析し、そして、その警報発生パターンに対して予
め定められた整理された手順に従って探索を行い、被疑
装置を特定する診断方式が採られていた。2. Description of the Related Art Conventionally, when various alarms are issued from an alarm device installed in a failure diagnosis target, which one of the devices constituting the failure diagnosis target is a device suspected of having a high possibility of failure. As a method of diagnosing whether or not there is a problem, pay attention to the relationship between a plurality of generated alarms, analyze the alarm occurrence patterns from among them, and search according to a predetermined organized procedure for the alarm occurrence patterns. The diagnostic method of identifying the suspected device has been adopted.
【0003】[0003]
【発明が解決しようとする課題】従来の故障診断方式で
は、以下のような課題がある。即ち、故障診断業務に携
わった専門的保守者から警報分析知識を聞き出して整理
する際に、その知識の中には、個々の警報とそれに対す
る被疑装置範囲との因果関係に関するものと、複数の警
報間の関連を分析し警報発生パターンを特定するもの
と、警報発生パターンとその被疑装置範囲との因果関係
に関するもの、などの複数種類の知識があるため、知識
の整理に多くの時間が必要であり、かつ、その後の知見
に基づく知識の修正も困難であった。The conventional failure diagnosis system has the following problems. That is, when hearing and organizing alarm analysis knowledge from a professional maintenance person who is involved in failure diagnosis work, some of the knowledge includes causal relationships between individual alarms and the range of suspected devices, and It takes a lot of time to organize the knowledge because there are multiple types of knowledge, such as analyzing the relationship between alarms to identify the alarm occurrence pattern and the causal relationship between the alarm occurrence pattern and the suspected device range. It was also difficult to correct the knowledge based on the subsequent knowledge.
【0004】また警報分析知識の中の、複数の警報間の
関連を分析し警報発生パターンを特定する知識と、警報
発生パターンとその被疑装置範囲との因果関係に関する
知識とを、経験的知識として獲得しているため、知識を
獲得する際に、保守者が故障診断業務を経験するのに必
要な一定の期間が必要であった。このため、診断対象シ
ステムのシステム開発と平行して故障診断エキスパート
システムを開発することができないという課題があっ
た。Further, among the alarm analysis knowledge, knowledge for analyzing a relation between a plurality of alarms to specify an alarm generation pattern and knowledge for a causal relationship between the alarm generation pattern and the suspected device range are used as empirical knowledge. Since they have acquired the knowledge, it took a certain period of time for the maintenance personnel to experience the failure diagnosis work when acquiring the knowledge. Therefore, there is a problem that the failure diagnosis expert system cannot be developed in parallel with the system development of the diagnosis target system.
【0005】更に、専門的保守者から得た警報分析知識
には、想定した警報発生パターンのみについての分析知
識しか記述されていないために、異なる警報発生パター
ンが発生した場合には、それに対応できないという課題
があった。Further, since the alarm analysis knowledge obtained from the professional maintenance staff only describes the analysis knowledge of only the expected alarm generation pattern, it cannot be dealt with when different alarm generation patterns occur. There was a problem.
【0006】本発明の目的は、上記従来の技術的課題を
克服して、専門的保守者からの警報分析知識の整理に多
くの時間を必要とせず、その追加修正も容易であり、そ
の上、診断対象システムのシステム開発と平行して故障
診断エキスパートシステムの開発が可能であり、如何な
る警報発生パターンにも対応可能である故障診断装置を
提供することにある。The object of the present invention is to overcome the above-mentioned conventional technical problems, to save a lot of time to arrange alarm analysis knowledge from a professional maintenance person, and to easily add and correct the alarm analysis knowledge. The purpose of the present invention is to provide a failure diagnosis device capable of developing a failure diagnosis expert system in parallel with the system development of a diagnosis target system and capable of coping with any alarm generation pattern.
【0007】[0007]
【課題を解決するための手段】上記目的達成のため、本
発明では、故障診断対象に設置された警報装置から各種
警報が発生した場合に、その警報を分析することによ
り、前記故障診断対象を構成する各装置の中で、どの装
置が故障の可能性が高く、どの装置が低いか、を判定す
る故障診断装置において、In order to achieve the above object, according to the present invention, when various alarms are issued from an alarm device installed in a failure diagnosis target, the alarms are analyzed to detect the failure diagnosis target. Of the devices that make up, a failure diagnosis device that determines which device has a high probability of failure and which device is low,
【0008】各種類別の警報と、それぞれの警報に対応
して故障の可能性のある装置を、その故障の可能性の程
度を表わす確信度と共に、対応付けて記憶する警報分析
知識テーブルと、An alarm analysis knowledge table in which each type of alarm and an apparatus having a possibility of failure corresponding to each type of alarm are associated and stored together with a certainty factor indicating the degree of the possibility of failure.
【0009】警報装置から発生した各種警報を受付ける
警報入力処理部と、前記警報入力処理部で受付けたそれ
ぞれの警報について、前記警報分析知識テーブルを参照
することにより、故障の可能性のある装置を、その故障
の可能性の程度である確信度と共に、拾いだしてきて、
整理する確信度設定処理部と、By referring to the alarm analysis knowledge table for the alarm input processing unit that receives various alarms generated from the alarm device and the respective alarms received by the alarm input processing unit, a device with a possibility of failure is identified. , With the certainty that is the degree of the possibility of failure, pick it up,
A certainty factor setting processing unit to be organized,
【0010】前記確信度設定処理部で整理された各警報
毎の、故障の可能性のある装置について、複数の警報に
またがっている装置については、各警報毎の確信度を合
成して確信度合成結果を算出する確信度合成処理部と、
前記確信度合成処理部で合成された各装置毎の確信度合
成結果を、しきい値と比較して、比較結果からそれぞれ
の装置の故障判定を行う故障判定処理部と、を具備し
た。With respect to the devices having a possibility of failure for each of the alarms arranged by the certainty factor setting processing unit, for the devices straddling a plurality of alarms, the certainty factors of the respective alarms are combined to generate the certainty factor. A certainty factor combination processing unit for calculating a combination result,
The certainty factor combination result for each device synthesized by the certainty factor combination processing unit is compared with a threshold value, and a failure determination processing unit for making a failure determination of each device from the comparison result is provided.
【0011】[0011]
【作用】複数の警報が発生した場合に、警報分析知識を
用いて、個々の警報に対して故障の可能性のある被疑装
置範囲とそれぞれの被疑装置の確信度を設定し、複数の
警報に対して重複して被疑装置となっている装置につい
ては、関係しているすべての警報に対する、それぞれの
確信度を計算式を用いて合成することにより、複数の警
報間の関連も考慮した複合的な警報分析を行う。When a plurality of alarms are generated, the alarm analysis knowledge is used to set the range of suspected devices having a possibility of failure and the certainty factor of each suspected device for each alarm, and the plurality of alarms are set. On the other hand, for devices that are duplicating suspected devices, by combining the certainty factors of all the related alarms using a formula, it is possible to combine multiple alarms into consideration. Alert analysis.
【0012】一方、正常値の限界を示す正常しきい値
と、異常値の限界を示す異常しきい値と、を設定してお
き、警報分析を行って確信度合成を行った後、その合成
確信度が正常しきい値以下の装置は正常と判定して被疑
範囲から取り除き、異常しきい値を越えた装置は、故障
装置と判定する。On the other hand, a normal threshold value indicating the limit of the normal value and an abnormal threshold value indicating the limit of the abnormal value are set, alarm analysis is performed, and certainty factor synthesis is performed. A device with a certainty factor equal to or lower than the normal threshold value is judged to be normal and removed from the suspected range, and a device exceeding the abnormal threshold value is judged to be a failed device.
【0013】[0013]
【実施例】次に図を参照して本発明の実施例を説明す
る。図1は、本発明の一実施例の全体的構成を示すブロ
ック図である。同図において、11は警報入力処理部、
12は確信度設定処理部、13は確信度合成処理部、1
4は故障判定部、21は警報分析知識を記憶するテーブ
ル、である。Embodiments of the present invention will now be described with reference to the drawings. FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention. In the figure, 11 is an alarm input processing unit,
12 is a confidence factor setting processing unit, 13 is a confidence factor combining processing unit, 1
Reference numeral 4 is a failure determination unit, and 21 is a table that stores alarm analysis knowledge.
【0014】動作の概要は次の如くである。警報入力処
理部11は、故障診断対象に設置された図示せざる警報
装置から発生した各種警報を受付ける。確信度設定処理
部12は、警報入力処理部11で受付けたそれぞれの警
報について、警報分析知識テーブル21を参照すること
により、故障の可能性のある装置を被疑装置として、そ
の故障の可能性の程度である確信度と共に、拾いだして
きて、整理する。The outline of the operation is as follows. The alarm input processing unit 11 receives various alarms generated from an alarm device (not shown) installed as a failure diagnosis target. The certainty factor setting processing unit 12 refers to the alarm analysis knowledge table 21 for each alarm received by the alarm input processing unit 11, and determines a device with a possibility of failure to be a device with a possibility of failure. With a certain degree of certainty, pick them up and organize them.
【0015】確信度合成処理部13は、確信度設定処理
部12で整理された各警報毎の、故障の可能性のある装
置(被疑装置)について、複数の警報にまたがっている
装置については、各警報毎の確信度を合成して確信度合
成結果を算出する。故障判定部14は、確信度合成処理
部13で合成された各被疑装置毎の確信度合成結果を、
しきい値と比較して、比較結果からそれぞれの被疑装置
の故障判定を行う。The certainty factor synthesis processing unit 13 determines, for each alarm sorted by the certainty factor setting processing unit 12, a device (suspected device) with a possibility of a failure, for a device that straddles a plurality of alarms. The certainty factor for each alarm is combined to calculate a certainty factor combination result. The failure determination unit 14 uses the certainty factor synthesis result for each suspected device, which is synthesized by the certainty factor synthesis processing unit 13,
By comparing with the threshold value, the failure judgment of each suspected device is made from the comparison result.
【0016】以下、図1における各ブロックについて、
更に詳しく説明する。図2は、図1における警報分析知
識テーブル21の具体例を示す説明図である。図2にお
いて、発生する警報の種別として、警報A、警報B、警
報Cがあることが分かる。Hereinafter, for each block in FIG.
This will be described in more detail. FIG. 2 is an explanatory diagram showing a specific example of the alarm analysis knowledge table 21 in FIG. In FIG. 2, it can be seen that the types of alarms that occur are alarm A, alarm B, and alarm C.
【0017】警報Aが発生すると、それにより故障の被
疑範囲に入る装置としては、装置1、装置2、装置3が
あり、装置1のそのときの故障の可能性の程度である確
信度は−0.2、装置2のそのときの故障の可能性の程
度である確信度は0.4、装置3のそのときの故障の可
能性の程度である確信度は0.3であることが分かる。
警報B、警報Cについても同様に、それが発生したとき
に故障の被疑範囲に入る装置と、その確信度が定義され
ている。これらの警報分析知識は、既知のものとしてテ
ーブル21に定義されているわけである。When the alarm A is generated, there are the devices 1, 2, and 3 as devices that fall within the suspicious range of the failure, and the certainty factor, which is the degree of the possibility of the failure of the device 1 at that time, is −. It can be seen that 0.2, the certainty factor that is the degree of possibility of failure of the device 2 is 0.4, and the certainty factor that is the degree of possibility of failure of the device 3 at that time is 0.3. .
Similarly, with respect to the alarms B and C, the confidence factor is defined as a device that falls within the suspicious range of the failure when it occurs. These alarm analysis knowledges are defined in the table 21 as known ones.
【0018】警報入力処理部11(図1)で警報A、警
報B、警報Cの3つの警報を受付けた場合、図2のテー
ブルに示す警報分析知識により、警報Aにより被疑範囲
に入ることになる装置とその確信度は、装置1(確信度
−0.2)、装置2(確信度0.4)、装置3(確信度
0.3)であり、警報Bにより被疑範囲に入ることにな
る装置とその確信度は、装置2(確信度−0.3)、装
置3(確信度0.4)であり、警報Cにより被疑範囲に
入ることになる装置とその確信度は、装置3(確信度
0.5)、装置4(確信度0.6)である。それでその
ように確信度設定処理部12で設定されるわけである。When the alarm input processing unit 11 (FIG. 1) receives three alarms A, B, and C, the alarm analysis knowledge shown in the table of FIG. And the certainty factors thereof are the device 1 (confidence factor −0.2), the device 2 (confidence factor 0.4), and the device 3 (confidence factor 0.3). The device and its certainty factor are the device 2 (certainty factor −0.3) and the device 3 (certainty factor 0.4), and the device and its certainty factor that will be in the suspicious range due to the alarm C are the device 3 and (Certainty factor 0.5) and device 4 (certainty factor 0.6). Therefore, the certainty factor setting processing unit 12 sets it as such.
【0019】図3は、図1における確信度合成処理部1
3で実行される確信度合成処理の説明図である。既に述
べたように、警報入力処理部11で警報A、警報B、警
報Cの3つの警報を受付けた場合、確信度設定処理部1
2で、図2に示す警報分析知識の通り、各警報により被
疑範囲に入ることになる装置とその確信度が設定され
る。FIG. 3 shows the certainty factor synthesis processing unit 1 in FIG.
6 is an explanatory diagram of a certainty factor combination process executed in FIG. As described above, when the alarm input processing unit 11 receives the three alarms A, B, and C, the certainty factor setting processing unit 1
In step 2, according to the alarm analysis knowledge shown in FIG. 2, the device and the certainty factor thereof which are to be in the suspected range by each alarm are set.
【0020】そこで改めて図2を見ると、装置2が警報
A、警報Bのそれぞれの被疑範囲に重なって入ってお
り、また、装置3が警報A、警報B、警報Cのそれぞれ
の被疑範囲に重なって入っていることが分かる。そし
て、モードが単一故障モードの場合(故障装置は一つに
限られている場合)、全ての警報に対して重複してそれ
ぞれの被疑範囲に入っている装置3についてのみ、その
確信度合成(それぞれの被疑範囲にある確信度の合成)
を行い、他の装置は全て確信度を0とする。また、複数
故障モードの場合(故障装置は一つではなく、複数ある
場合)、一つ以上の警報に対して被疑範囲に入っている
装置1と装置2と装置3と装置4の、それぞれについて
確信度合成を行う。Referring again to FIG. 2, the device 2 is included in the suspicious range of the alarm A and the alarm B, and the device 3 is included in the suspicious range of the alarm A, the alarm B, and the alarm C. You can see that they overlap. Then, when the mode is the single failure mode (when the number of failed devices is limited to one), the confidence factor combination is performed only for the devices 3 that are redundantly included in the respective suspicious ranges for all the alarms. (Synthesis of certainty factors within each suspected range)
And other devices set the certainty factor to 0. Further, in the case of the multiple failure mode (when there are a plurality of failure devices instead of one), the device 1, the device 2, the device 3, and the device 4, which are in the suspicious range for one or more alarms, respectively. Confidence factor synthesis is performed.
【0021】かかる確信度合成が確信度合成処理部13
で行われ、その結果が図3に示されているわけである。
以下、図3を参照しながら、確信度を合成する計算処理
について説明する。The certainty factor synthesis is performed by the certainty factor synthesis processing unit 13.
The result is shown in FIG.
Hereinafter, the calculation process for combining the certainty factors will be described with reference to FIG.
【0022】警報A、警報Bのそれぞれの被疑範囲に重
なって入った装置2の合成確信度を求める場合、図3に
示すように、警報A(証拠e1)に基づく装置2の故障
仮設(仮設h2)の確信度をCF(h2/e1)とし、
警報B(証拠e2)に基づく装置2の故障仮設(仮設h
2)の確信度をCF(h2/e2)とすると、警報A
(証拠e1)と警報B(証拠2)に基づく装置2の故障
仮設(仮設h2)の合成確信度CF(h2/e1,e
2)は、以下の計算方法で求められる。In order to obtain the combined certainty factor of the device 2 that is included in the suspicious ranges of the alarm A and the alarm B, as shown in FIG. 3, the failure temporary (temporary installation) of the device 2 based on the alarm A (evidence e1) is performed. The confidence factor of h2) is CF (h2 / e1),
Temporary failure of device 2 based on alarm B (evidence e2) (temporary h
If the certainty factor of 2) is CF (h2 / e2), alarm A
Combined confidence factor CF (h2 / e1, e) of the failure temporary (temporary h2) of the device 2 based on (evidence e1) and alarm B (evidence 2)
2) is calculated by the following calculation method.
【0023】ここでの考え方は、警報の被疑範囲として
重なった数の多いものほど、故障の疑いが高くなる。つ
まり、確信度が高くなる。さらに、警報の重なりのない
高い確信度のものよりも、低い確信度ではあるが、数多
く重なっているものの方がより故障の確信度が高くな
り、確信度が負の場合は、負の方に高くなると考えてい
る。よって、複数警報が発生した場合に、重複して被疑
範囲に入っている装置については、関係している全ての
警報に対する確信度を、計算式を用いて合成する手法を
図1の確信度合成処理部13に組み込んであるわけであ
る。The idea here is that the larger the number of overlapping alarm suspect ranges, the higher the suspicion of failure. That is, the degree of certainty increases. Furthermore, although the confidence is lower than that of the high confidence without overlapping of alarms, the confidence of the failure is higher when there are many overlaps, and when the confidence is negative, the confidence is negative. I think it will be higher. Therefore, when a plurality of alarms are generated, for a device that is in the suspected range in duplicate, the method of combining the certainty factors for all the related alarms using a calculation formula is used as the certainty factor combination of FIG. It is incorporated in the processing unit 13.
【0024】さて計算方法であるが、以下に示す計算式
A−1〜A−3は、MYCINなどで用いられている確
証理論の確信度結合規則による式である。警報の被疑範
囲が重なった場合の確信度合成計算式は、かかる計算式
A−1〜A−3であり、以下具体的に説明する。Now, regarding the calculation method, the following calculation formulas A-1 to A-3 are formulas based on the certainty factor combination rule of the confirmation theory used in MYCIN and the like. The certainty factor combination calculation formulas when the suspicious ranges of the alarms overlap are the calculation formulas A-1 to A-3, which will be specifically described below.
【0025】(計算式A−1)警報e1と警報e2が発
生した場合に、警報e1に基づく装置hの故障仮設の確
信度が正の場合と、警報e2に基づく装置hの故障仮設
の確信度が正の場合の、症状が重なった場合には、装置
hが故障である確率が高いと判断し、確信度を高くする
ための計算式を用いる。即ち(Calculation formula A-1) When the alarm e1 and the alarm e2 occur, the certainty factor of the temporary failure of the device h based on the alarm e1 is positive, and the certainty of the temporary failure of the device h based on the alarm e2. When the degree is positive and the symptoms overlap, it is determined that the probability that the device h is out of order is high, and a calculation formula for increasing the certainty is used. I.e.
【0026】 計算式A−1: if CF(h,e1 )>0 , CF(h,e2 )>0 then CF(h,e1 ,e2 )= CF(h,e1 )+CF(h,e2 )−CF(h,e1 )・CF(h,e2 )Calculation Formula A-1: if CF (h, e 1 )> 0, CF (h, e 2 )> 0 then CF (h, e 1 , e 2 ) = CF (h, e 1 ) + CF ( h, e 2) -CF (h , e 1) · CF (h, e 2)
【0027】(計算式A−2)警報e1と警報e2が発
生した場合に、警報e1に基づく装置hの故障仮設の確
信度が負の場合と、警報e2に基づく装置hの故障仮設
の確信度が負の場合の、症状が重なった場合には、装置
hが故障である確率が低いと判断し、確信度を低くする
ための計算式を用いる。即ち(Calculation formula A-2) When the alarm e1 and the alarm e2 are generated, the certainty factor of the temporary failure of the device h based on the alarm e1 is negative, and the certainty factor of the temporary failure of the device h based on the alarm e2. When the degree is negative and the symptoms overlap, it is determined that the probability that the device h is out of order is low, and a calculation formula for lowering the certainty factor is used. I.e.
【0028】 計算式A−2: if CF(h,e1 )<0 ,CF(h,e2 )<0 then CF(h,e1 ,e2 )= CF(h,e1 )+CF(h,e2 )+CF(h,e1 )・CF(h,e2 )Formula A-2: if CF (h, e 1 ) <0, CF (h, e 2 ) <0 then CF (h, e 1 , e 2 ) = CF (h, e 1 ) + CF ( h, e 2 ) + CF (h, e 1 ) · CF (h, e 2 ).
【0029】(計算式A−3)警報e1と警報e2が発
生した場合に、警報e1に基づく装置hの故障仮設の確
信度が正の場合と、警報e2に基づく装置hの故障仮設
の確信度が負の場合の、症状が重なった場合には、装置
hの確信度を更新するための計算式として、次に示す数
1式を用いる。即ち(Calculation formula A-3) When the alarm e1 and the alarm e2 are generated, the certainty factor of the temporary failure of the device h based on the alarm e1 is positive, and the certainty of the temporary failure of the device h based on the alarm e2. When the degree is negative and the symptoms overlap, the following formula 1 is used as a calculation formula for updating the certainty factor of the device h. I.e.
【0030】 計算式A−3: if CF(h,e1 )>0 , CF(h,e2 )<0 thenCalculation Formula A-3: if CF (h, e 1 )> 0, CF (h, e 2 ) <0 then
【0031】[0031]
【数1】 [Equation 1]
【0032】上記計算式A−1〜A−3では低い確信度
同士の合成では、合成結果としてあまり高い確信度には
ならない。そこで、低い確信度同士でも、合成結果が高
い確信度になるように、計算式B−1〜B−4が考案さ
れている。In the above calculation formulas A-1 to A-3, the combination of low certainty factors does not result in a very high certainty factor. Therefore, calculation formulas B-1 to B-4 have been devised so that the combined result has a high certainty factor even if the certainty factors are low.
【0033】確証理論の確信度結合規則による式である
計算式A−1〜A−3は、各変数に確信度をそのまま用
いていたが、計算式B−1〜B−4は、確信度が−1.
0〜+1.0であることから、確信度の絶対値をとり
0.0〜1.0の数値に変換して、0.0〜1.0の数
値では、√平方根をとったものは、元の数値よりも常に
高くなる。しかも、その高くなる度合いは、元の数値が
小さいほど大きくなるという性質を用いて、各変数に確
信度の√平方根をとり、それに元の確信度の正負の符号
を付けることにより、計算式A−1〜A−3よりも、確
信度の合成値が高くなるように考案された計算式であ
る。In the calculation formulas A-1 to A-3, which are the formulas based on the certainty factor combination rule of the confirmation theory, the certainty factor is used as it is for each variable, but the calculation formulas B-1 to B-4 are the certainty factors. Is -1.
Since it is 0 to +1.0, the absolute value of the certainty factor is taken and converted into a numerical value of 0.0 to 1.0, and the numerical value of 0.0 to 1.0 takes the square root of It will always be higher than the original number. Moreover, the degree of increase becomes larger as the original numerical value becomes smaller, and the square root of the certainty factor is taken for each variable, and the sign of the original certainty factor is added to the calculation formula A It is a calculation formula devised so that the combined value of the certainty factors is higher than those of -1 to A-3.
【0034】計算式B−1〜B−4の考え方は、前記計
算式A−1〜A−3の考え方と同様である。 計算式B−1: if CF(h,e1 )>0, CF(h,e2 )>0 then CF(h,e1 ,e2 )= SQRT(CF(h,e1 ))+SQRT(CF(h,e2 )) −SQRT(CF(h,e1 )・CF(h,e2 ))The idea of the calculation formulas B-1 to B-4 is the same as that of the calculation formulas A-1 to A-3. Formula B-1: if CF (h , e 1)> 0, CF (h, e 2)> 0 then CF (h, e 1, e 2) = SQRT (CF (h, e 1)) + SQRT ( CF (h, e 2 ))-SQRT (CF (h, e 1 ) CF (h, e 2 ))
【0035】 計算式B−2: if CF(h,e1 )<0 CF(h,e2 )<0 then CF(h,e1 ,e2 )= −SQRT(ABS(CF(h,e1 ))− SQRT(ABS(CF(h,e2 ))+ SQRT(CF(h,e1 )・CF(h,e2 ))Formula B-2: if CF (h, e 1 ) <0 CF (h, e 2 ) <0 then CF (h, e 1 , e 2 ) = − SQRT (ABS (CF (h, e 1)) - SQRT (ABS ( CF (h, e 2)) + SQRT (CF (h, e 1) · CF (h, e 2))
【0036】 計算式B−3: if CF(h,e1 )>0, CF(h,e2 )<0 thenFormula B-3: if CF (h, e 1 )> 0, CF (h, e 2 ) <0 then
【0037】[0037]
【数2】 [Equation 2]
【0038】 計算式B−4 : if CF(h,e1 )<0, CF(h,e2 )>0 thenFormula B-4: if CF (h, e 1 ) <0, CF (h, e 2 )> 0 then
【0039】[0039]
【数3】 [Equation 3]
【0040】但し、CF:確信度(Certainty
Factors)、その確信度の値の範囲は(−1.
0〜+1.0) 、CF(h,e):証拠eに基づく仮
設hの確信度(証拠eに基づいて仮設hを支持するか、
または支持しないかの度合い)、h:仮設(hypothesi
s)、e:証拠(evidence)、である。However, CF: Certainty (Certainty)
Factors), and the range of the value of the certainty factor is (-1.
0 to +1.0), CF (h, e): confidence of the temporary h based on the evidence e (whether the temporary h is supported based on the evidence e,
Or, whether it is not supported), h: hypothesi
s), e: evidence.
【0041】以上、計算式について説明した。そこで、
図3において、警報Aと警報Bに関して、装置2の確信
度を合成する場合、警報Aに基づく確信度と、警報Bに
基づく確信度を、以下のように合成する。The calculation formula has been described above. Therefore,
In FIG. 3, when the certainty factors of the device 2 are combined with respect to the alarm A and the alarm B, the certainty factor based on the alarm A and the certainty factor based on the alarm B are combined as follows.
【0042】警報Aに基づく確信度CF(h2 ,e1 )
=0.4 >0、 警報Bに基づく確信度CF(h2 ,e2 )=−0.3
<0、 であるので上記計算式B−3を用いて、Confidence factor CF (h 2 , e 1 ) based on alarm A
= 0.4> 0, confidence CF (h 2, e 2) based on the alarm B = - 0.3
<0, so using the above calculation formula B-3,
【0043】[0043]
【数4】 となる。[Equation 4] Becomes
【0044】次に、図3における装置3(仮設h3)の
ように、2つ以上の確信度を合成する場合は、2つの確
信度を合成する計算を繰り返すことにより以下の計算方
法で合成確信度が求められる。Next, in the case of combining two or more certainty factors like the device 3 (temporary h3) in FIG. 3, by repeating the calculation of combining two certainty factors, the combined certainty is calculated by the following calculation method. Degree is required.
【0045】警報Aと警報Bと警報Cに関して、装置3
の確信度を合成する場合、最初に、警報Aに基づく確信
度と、警報Bに基づく確信度を、以下のように合成す
る。 警報Aに基づく確信度CF(h3 ,e1 )=0.3 >
0、 警報Bに基づく確信度CF(h3 ,e2 )=0.4 >
0、 であるので計算式B−1を用いて、Regarding the alarm A, the alarm B, and the alarm C, the device 3
In the case of synthesizing the certainty factors, the certainty factors based on the alarm A and the certainty factors based on the alarm B are first synthesized as follows. Confidence factor CF (h 3 , e 1 ) based on the alarm A = 0.3>
0, confidence factor CF (h 3 , e 2 ) based on alarm B = 0.4>
0, so using the calculation formula B-1,
【0046】CF(h2 ,e1 ,e2 )=SQRT(C
F(h3 ,e1 ))+SQRT(CF(h3 ,e2 ))
−SQRT(CF(h3 ,e1 )・CF(h3 ,e2 ) =0.83 となる。CF (h 2 , e 1 , e 2 ) = SQRT (C
F (h 3 , e 1 )) + SQRT (CF (h 3 , e 2 ))
−SQRT (CF (h 3 , e 1 ) · CF (h 3 , e 2 ) = 0.83.
【0047】次に、上で求めた合成確信度と警報Cに基
づく確信度を以下のように合成する。 警報Aと警報Bに基づく合成確信度CF(h3 ,e1 ,
e2 )=0.83 >0、 警報Cに基づく確信度CF(h3 ,e3 )=0.5 >
0、であるので計算式B−1を用いて、Next, the combined certainty factor obtained above and the certainty factor based on the alarm C are combined as follows. Combined confidence factor CF (h 3 , e 1 , based on alarm A and alarm B,
e 2 ) = 0.83> 0, confidence factor CF (h 3 , e 3 ) = 0.5> based on the alarm C
Since it is 0, using the calculation formula B-1,
【0048】CF(h3 ,e1 ,e2 ,e3 )=SQR
T(h3 ,e1 ,e2)+SQRT(CF(h3 ,
e3 ))−SQRT(CF(h3 ,e1 ,e2 )・CF
(h3 ,e3 ) =0.97 となる。CF (h 3 , e 1 , e 2 , e 3 ) = SQR
T (h 3 , e 1 , e 2) + SQRT (CF (h 3 ,
e 3)) - SQRT (CF (h 3, e 1, e 2) · CF
(H 3 , e 3 ) = 0.97.
【0049】以上で確信度合成処理部13の説明を終
え、故障判定処理部14について説明する。確信度を合
成した結果、図3の(確信度合成結果)の欄に見られる
ように、装置1の確信度が−0.2、装置2の確信度が
0.19、装置3の確信度が0.97、装置4の確信度
が0.6となった。この場合、図4に見られるように、
正常しきい値(正常の限界となるしきい値)が0.0、
異常しきい値(異常の限界となるしきい値)が0.9で
あれば、故障判定処理部14は、確信度が正常しきい値
以下の装置である装置1を、正常な装置と判定し、異常
しきい値を越えた装置である装置3を故障装置と判定す
る。The description of the certainty factor synthesis processing unit 13 is completed above, and the failure determination processing unit 14 will be described. As a result of combining the certainty factors, the certainty factor of the device 1 is −0.2, the certainty factor of the device 2 is 0.19, and the certainty factor of the device 3 is as shown in the column of (Confidence factor combination result) in FIG. Was 0.97, and the confidence factor of the device 4 was 0.6. In this case, as seen in FIG.
Normal threshold (threshold that is the limit of normality) is 0.0,
If the abnormal threshold value (threshold value that is the limit of abnormality) is 0.9, the failure determination processing unit 14 determines that the device 1 having a certainty factor equal to or lower than the normal threshold value is a normal device. Then, the device 3, which is a device that exceeds the abnormal threshold value, is determined to be a failed device.
【0050】[0050]
【発明の効果】以上説明したように、本発明によれば、
故障診断システムにおける警報分析処理を構築する際、
警報分析知識として複数の警報間の関連を分析して警報
発生パターンを特定する知識と、警報発生パターンとそ
の対応した被疑範囲との因果関係に関する知識を整理す
る必要がなく、個々の警報とその対応した被疑範囲との
因果関係に関するもののみでよいために、知識の獲得及
び、知識の修正が容易となる。As described above, according to the present invention,
When constructing an alarm analysis process in a failure diagnosis system,
It is not necessary to organize knowledge that analyzes the relationship between multiple alarms to identify the alarm occurrence pattern as alarm analysis knowledge and knowledge about the causal relationship between the alarm occurrence pattern and the corresponding suspected range, and Since only the causal relationship with the corresponding suspicious range is required, knowledge acquisition and knowledge correction are facilitated.
【0051】また、警報分析知識を警報設定時の設計者
の知識として警報発生のメカニズム知識を整理すること
により、専門的保守者からの経験的知識を不要とするこ
とで、故障診断業務を経験するのに必要な一定の期間を
不要とし、診断対象システムのシステム開発と平行して
故障診断エキスパートシステムを開発することが可能と
なる。Further, by organizing the alarm generation mechanism knowledge as the alarm analysis knowledge as the designer's knowledge at the time of alarm setting, the experience of a specialist maintenance person becomes unnecessary, and the failure diagnosis work is experienced. It becomes possible to develop a fault diagnosis expert system in parallel with the system development of the system to be diagnosed by eliminating the fixed period required for the operation.
【0052】さらに、複数の装置から複数の警報が発生
した場合に、複数の警報にまたがって被疑範囲に入って
いる装置については、関係している全ての警報に対する
確信度を合成することにより、複数の警報間の関連も考
慮した複合的な警報分析が可能となる。Further, when a plurality of alarms are generated from a plurality of devices, for devices that are in the suspected range over a plurality of alarms, by combining the certainty factors for all the related alarms, It is possible to perform a complex alarm analysis that considers the relationship between multiple alarms.
【図1】本発明の一実施例の全体的構成を示すブロック
図である。FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention.
【図2】図1における警報分析知識テーブル21の具体
例を示す説明図である。FIG. 2 is an explanatory diagram showing a specific example of an alarm analysis knowledge table 21 in FIG.
【図3】図1における確信度合成処理部13で実行され
る確信度合成処理の説明図である。3 is an explanatory diagram of a certainty factor combination process executed by a certainty factor combination processing unit 13 in FIG. 1. FIG.
【図4】図1における故障判定処理部14で実行される
故障判定処理の説明図である。4 is an explanatory diagram of a failure determination process executed by a failure determination processing unit 14 in FIG.
11…警報入力処理部、12…確信度設定処理部、13
…確信度合成処理部、14…故障判定処理部、21…警
報分析知識テーブル11 ... Alarm input processing unit, 12 ... Confidence factor setting processing unit, 13
... certainty factor synthesis processing unit, 14 ... failure determination processing unit, 21 ... alarm analysis knowledge table
───────────────────────────────────────────────────── フロントページの続き (72)発明者 藤田 勝美 東京都千代田区内幸町1丁目1番6号 日 本電信電話株式会社内 (72)発明者 村里 由起夫 神奈川県横浜市中区山下町223番1号 エ ヌ・ティ・ティ・ソフトウェア株式会社内 (72)発明者 長末 彰 神奈川県横浜市中区山下町223番1号 エ ヌ・ティ・ティ・ソフトウェア株式会社内 (72)発明者 張 宏 神奈川県横浜市中区山下町223番1号 エ ヌ・ティ・ティ・ソフトウェア株式会社内 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Katsumi Fujita 1-1-6 Uchisaiwaicho, Chiyoda-ku, Tokyo Nihon Telegraph and Telephone Corporation (72) Yukio Murasato 223-1 Yamashita-cho, Naka-ku, Yokohama-shi, Kanagawa No. NTT Software Corporation (72) Inventor Akira Nagasue No. 223-1, Yamashita-cho, Naka-ku, Yokohama-shi, Kanagawa Prefecture NTT Software Corporation (72) Inventor Hiroshi Zhang Inside NTT Software Corporation, 223-1, Yamashita-cho, Naka-ku, Yokohama-shi, Kanagawa
Claims (1)
各種警報が発生した場合に、その警報を分析することに
より、前記故障診断対象を構成する各装置の中で、どの
装置が故障の可能性が高く、どの装置が低いか、を判定
する故障診断装置において、 各種類別の警報と、それぞれの警報に対応して故障の可
能性のある装置を、その故障の可能性の程度を表わす確
信度と共に、対応付けて記憶する警報分析知識テーブル
と、 警報装置から発生した各種警報を受付ける警報入力処理
部と、 前記警報入力処理部で受付けたそれぞれの警報につい
て、前記警報分析知識テーブルを参照することにより、
故障の可能性のある装置を、その故障の可能性の程度で
ある確信度と共に、拾いだしてきて、整理する確信度設
定処理部と、 前記確信度設定処理部で整理された各警報毎の、故障の
可能性のある装置について、複数の警報にまたがってい
る装置については、各警報毎の確信度を合成して確信度
合成結果を算出する確信度合成処理部と、 前記確信度合成処理部で合成された各装置毎の確信度合
成結果を、しきい値と比較して、比較結果からそれぞれ
の装置の故障判定を行う故障判定処理部と、を具備して
成ることを特徴とする故障診断装置。1. When various alarms are generated from an alarm device installed in a failure diagnosis target, by analyzing the alarms, which device among the devices constituting the failure diagnosis target is likely to have a failure. In a fault diagnosis device that determines which device has a high probability and which device has a low probability, each type of alarm and the device with a possibility of failure corresponding to each alarm are represented as a degree of probability of failure. The alarm analysis knowledge table stored in association with each other, the alarm input processing unit for receiving various alarms generated from the alarm device, and the alarm analysis knowledge table for each alarm received by the alarm input processing unit. By
A device with a possibility of failure, together with a certainty factor which is the degree of the possibility of the failure, is picked up and arranged, and a certainty factor setting processing unit, and for each alarm arranged by the certainty factor setting processing unit. A device with a possibility of failure, and a device that straddles a plurality of alarms, a certainty factor synthesis processing unit that synthesizes certainty factors for each alarm to calculate a certainty factor synthesis result; The certainty factor synthesis result for each device synthesized by the unit is compared with a threshold value, and a failure determination processing unit for performing a failure determination of each device based on the comparison result is provided. Fault diagnosis device.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP5257971A JPH07114483A (en) | 1993-10-15 | 1993-10-15 | Fault diagnosing device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP5257971A JPH07114483A (en) | 1993-10-15 | 1993-10-15 | Fault diagnosing device |
Publications (1)
Publication Number | Publication Date |
---|---|
JPH07114483A true JPH07114483A (en) | 1995-05-02 |
Family
ID=17313761
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
JP5257971A Pending JPH07114483A (en) | 1993-10-15 | 1993-10-15 | Fault diagnosing device |
Country Status (1)
Country | Link |
---|---|
JP (1) | JPH07114483A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007007410A1 (en) * | 2005-07-14 | 2007-01-18 | Fujitsu Limited | Message analyzing device, message analyzing method and message analyzing program |
JP2010181212A (en) * | 2009-02-04 | 2010-08-19 | Toyota Central R&D Labs Inc | System and method of diagnosing fault |
-
1993
- 1993-10-15 JP JP5257971A patent/JPH07114483A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007007410A1 (en) * | 2005-07-14 | 2007-01-18 | Fujitsu Limited | Message analyzing device, message analyzing method and message analyzing program |
JPWO2007007410A1 (en) * | 2005-07-14 | 2009-01-29 | 富士通株式会社 | Message analysis apparatus, control method, and control program |
US7823016B2 (en) | 2005-07-14 | 2010-10-26 | Fujitsu Limited | Message analyzing apparatus, message analyzing method, and computer product |
JP2010181212A (en) * | 2009-02-04 | 2010-08-19 | Toyota Central R&D Labs Inc | System and method of diagnosing fault |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5463768A (en) | Method and system for analyzing error logs for diagnostics | |
AU1103401A (en) | Method and apparatus for diagnosing difficult to diagnose faults in a complex system | |
EP1279211B1 (en) | Topology-based reasoning apparatus for root-cause analysis of network faults | |
CN105325023B (en) | Method and the network equipment for cell abnormality detection | |
EP0482526A2 (en) | Optimization method for adaptive sensor reading scheduling and delayed alarm evaluation in real-time diagnostic systems | |
US20040010733A1 (en) | System and method for fault identification in an electronic system based on context-based alarm analysis | |
JPH01243135A (en) | Problem processing system | |
AU2004201704A1 (en) | System and method for processing a new diagnostics case relative to historical case data and determining a ranking for possible repairs | |
JP2004118839A (en) | Method for supporting specification of function unit failed in technical equipment | |
EP2410312A1 (en) | A method for computer-assisted analyzing of a technical system | |
KR102073810B1 (en) | Method and system for predicting the failure of naval ship propulsion system using machine learning | |
CN109670610A (en) | Fault diagnosis optimization method based on fault propagation analysis | |
CN106646068A (en) | Method for diagnosing defects of intelligent substation secondary system based on multi-parameter information fusion | |
CN114020581A (en) | Alarm correlation method based on topological optimization FP-Growth algorithm | |
CN115145751A (en) | Method, device, equipment and storage medium for positioning fault root cause of micro-service system | |
CN113242213B (en) | Power communication backbone network node vulnerability diagnosis method | |
CN112882875A (en) | Fault diagnosis method | |
JPH07114483A (en) | Fault diagnosing device | |
CN108825482A (en) | A kind of fault detection method and detection system of aircraft axial plunger pump | |
CN110609761B (en) | Method and device for determining fault source, storage medium and electronic equipment | |
CN113328898B (en) | Fault diagnosis method and system with autonomous learning capability | |
CN110059413A (en) | A kind of method for diagnosing faults | |
CN114629776B (en) | Fault analysis method and device based on graph model | |
DePold et al. | Metrics for evaluating the accuracy of diagnostic fault detection systems | |
JPH022406A (en) | Device for fault diagnosis of plant |