JP4832609B1 - Abnormal sign diagnosis device and abnormality sign diagnosis method - Google Patents
Abnormal sign diagnosis device and abnormality sign diagnosis method Download PDFInfo
- Publication number
- JP4832609B1 JP4832609B1 JP2011138941A JP2011138941A JP4832609B1 JP 4832609 B1 JP4832609 B1 JP 4832609B1 JP 2011138941 A JP2011138941 A JP 2011138941A JP 2011138941 A JP2011138941 A JP 2011138941A JP 4832609 B1 JP4832609 B1 JP 4832609B1
- Authority
- JP
- Japan
- Prior art keywords
- diagnosis
- abnormality
- sensor data
- data
- sign
- 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.)
- Active
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 330
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 54
- 230000000875 corresponding Effects 0.000 claims description 26
- 230000005484 gravity Effects 0.000 claims description 5
- 238000002405 diagnostic procedure Methods 0.000 claims description 4
- 238000007418 data mining Methods 0.000 description 41
- 238000005259 measurement Methods 0.000 description 26
- 238000004364 calculation method Methods 0.000 description 19
- 238000004891 communication Methods 0.000 description 15
- 238000000034 method Methods 0.000 description 10
- 239000000498 cooling water Substances 0.000 description 5
- 230000003287 optical Effects 0.000 description 5
- 239000004065 semiconductor Substances 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 230000001052 transient Effects 0.000 description 2
- 230000005526 G1 to G0 transition Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000001186 cumulative Effects 0.000 description 1
- 230000002354 daily Effects 0.000 description 1
- 230000003203 everyday Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000004171 remote diagnosis Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
Images
Abstract
【解決手段】異常予兆診断装置1は、機械設備2に設置された複数のセンサによって測定された多次元センサデータを取得するセンサデータ取得手段11と、診断対象データについて、機械設備2が正常に稼動しているときのセンサデータを学習データとして作成された事例モデルからの乖離の度合いを示す異常度に基づいて異常予兆の有無を診断する第1診断手段16と、個別のセンサデータの値が、予め定められた所定の範囲内にあるか否かに基づいて、異常予兆の有無を診断する第2診断手段15とを備え、第1診断手段16によって異常予兆があると診断された場合に、第2診断手段15による異常予兆診断のために参照する個別のセンサデータを、異常度に対する寄与の大きさを示す寄与度に基づいて選択する。
【選択図】図1An abnormality sign diagnosis apparatus and an abnormality sign diagnosis method that can appropriately diagnose whether or not there is an abnormality sign.
An abnormality sign diagnosis apparatus includes a sensor data acquisition unit for acquiring multidimensional sensor data measured by a plurality of sensors installed in a machine facility, and the machine facility is normally used for diagnosis target data. First diagnosis means 16 for diagnosing the presence / absence of an abnormality sign based on an abnormality degree indicating a degree of deviation from a case model created by using sensor data as learning data at the time of operation, and values of individual sensor data The second diagnostic means 15 for diagnosing the presence or absence of an abnormal sign based on whether or not the predetermined sign is within a predetermined range, and when the first diagnostic means 16 diagnoses that there is an abnormal sign Individual sensor data to be referred to for abnormality sign diagnosis by the second diagnosis means 15 is selected based on the contribution indicating the magnitude of the contribution to the abnormality.
[Selection] Figure 1
Description
[異常予兆診断装置の構成]
まず、図1を参照して、本発明の実施形態に係る異常予兆診断装置の構成について説明する。 Embodiments of the present invention will be described below with reference to the drawings.
[Configuration of abnormal sign diagnosis device]
First, with reference to FIG. 1, the structure of the abnormality sign diagnostic apparatus which concerns on embodiment of this invention is demonstrated.
また、通信手段10は、機械設備2から入力したセンサデータをセンサデータ取得手段11に出力し、機械設備2から入力した保守情報を保守情報取得手段13に出力する。 The communication means 10 is for communicating with the machine facility 2 via the communication network N. The communication means 10 may communicate with the machine facility 2 via a LAN (Local Area Network) or WAN (Wide Area Network), or may communicate directly with the machine facility 2 via a telephone line. Further, when inputting maintenance information such as a maintenance plan and a maintenance history of the mechanical equipment 2 in addition to the sensor data of the mechanical equipment 2 as information input from the mechanical equipment 2, even if the communication path differs depending on the input information. Good. In this case, what is necessary is just to provide several different communication means in order to acquire each information.
The communication unit 10 outputs the sensor data input from the machine facility 2 to the sensor data acquisition unit 11, and outputs the maintenance information input from the machine facility 2 to the maintenance information acquisition unit 13.
図2に示すように、本実施形態のリモートモニタリング部15は、多次元(N次元;Nは2以上の整数)のセンサデータを構成するN個の個別(一次元)のセンサデータのそれぞれに対応した個別診断手段15111〜1512Nと、総合診断手段1521〜1522とを備えている。なお、個別(一次元)のセンサデータとは、機械設備2における「冷却水圧力」、「冷却水温度」、「回転速度」などの、個々のセンサによる測定データのことであり、多次元のセンサデータとは、これらの個別のセンサデータを要素とするセンサデータのことである。 Here, a detailed configuration of the remote monitoring unit 15 will be described with reference to FIG.
As shown in FIG. 2, the remote monitoring unit 15 of the present embodiment applies each of N individual (one-dimensional) sensor data constituting multidimensional (N dimensions; N is an integer of 2 or more) sensor data. Corresponding individual diagnosis means 151 11 to 151 2N and comprehensive diagnosis means 152 1 to 152 2 are provided. The individual (one-dimensional) sensor data refers to data measured by individual sensors such as “cooling water pressure”, “cooling water temperature”, “rotational speed”, etc. in the mechanical equipment 2. The sensor data is sensor data having these individual sensor data as elements.
また、同じセンサデータ(センサ1、センサ2など)に対応する個別診断手段15111と個別診断手段15121、個別診断手段15112と個別診断手段15122、個別診断手段1511Nと個別診断手段1512Nなどは、それぞれ、同じ閾値を用いて個別診断するようにしてもよく、組となる総合診断手段1521〜1522が診断すべき異常種別に応じて、異なる閾値を用いるようにしてもよい。また、総合診断手段1521〜1522は、診断すべき異常種別に応じて、それぞれ参照するセンサデータを選別するようにしてもよい。 The individual diagnosis means 151 11 to 151 1N, the comprehensive diagnosis means 152 1 , the individual diagnosis means 151 21 to 151 2N, and the comprehensive diagnosis means 152 2 are each set as a set, and whether or not there is an abnormality sign corresponding to each abnormality type Is to diagnose. The number of sets is not particularly limited, and may be two or more or one.
Also, individual diagnosis means 151 11 and individual diagnosis means 151 21 , individual diagnosis means 151 12 and individual diagnosis means 151 22 , individual diagnosis means 151 1 N and individual diagnosis means 151 corresponding to the same sensor data (sensor 1, sensor 2 etc.) etc. 2N, respectively, using the same threshold may be individually diagnosed, depending on the overall diagnosis means 152 1 to 152 2 abnormality to be diagnosed type to be set may be used with different thresholds . Further, the comprehensive diagnosis means 152 1 to 152 2 may select sensor data to be referred to according to the abnormality type to be diagnosed.
なお、総合診断手段1521〜1522における診断ルールは、データマイニング部16の診断結果に関わらず、キーボードやマウスなどの入力手段(不図示)を介した操作者の指示があった場合に、再構築するようにしてもよい。 Further, when it is diagnosed that there is an abnormal sign by the abnormal sign diagnosis by the data mining unit 16, the comprehensive diagnosis units 152 1 to 152 2 , based on the contribution calculated by the data mining unit 16 at that time, Reconstruct diagnosis rules using multiple individual diagnosis results.
Note that the diagnosis rules in the comprehensive diagnosis means 152 1 to 152 2 are based on an operator's instruction via an input means (not shown) such as a keyboard or a mouse regardless of the diagnosis result of the data mining unit 16. You may make it rebuild.
データマイニング部(第1診断手段)16は、機械設備2から取得した多次元のセンサデータを参照して、統計的処理を施して事例モデルを作成し、作成した事例モデルを用いたデータマイニング手法により異常予兆の診断を行うものである。なお、本実施形態におけるデータマイニング部16は、センサデータ記憶手段12に記憶されているセンサデータを参照して診断する。また、データマイニング部16は、診断結果を機械設備2に対応付けて、診断したセンサデータの測定時刻とともに診断結果記憶手段18に記憶する。なお、本実施形態においては、データマイニング部16は、診断結果として異常予兆の有無に加えて、寄与度および異常度を含めて診断結果記憶手段18に記憶する。 Returning to FIG. 1, the description of the configuration of the abnormality sign diagnosis apparatus 1 will be continued.
The data mining unit (first diagnosis means) 16 refers to the multidimensional sensor data acquired from the mechanical equipment 2 and creates a case model by performing statistical processing, and a data mining method using the created case model Is used to diagnose abnormal signs. The data mining unit 16 in the present embodiment makes a diagnosis with reference to the sensor data stored in the sensor data storage unit 12. Further, the data mining unit 16 stores the diagnosis result in the diagnosis result storage unit 18 together with the measurement time of the diagnosed sensor data in association with the mechanical equipment 2. In the present embodiment, the data mining unit 16 stores the contribution result and the abnormality degree in the diagnosis result storage unit 18 in addition to the presence / absence of the abnormality sign as the diagnosis result.
図3に示すように、本実施形態のデータマイニング部16は、学習部161と診断部162とを備えている。また、学習部161は、学習データ取得手段161aと、モード分割手段161bと、モデル作成手段161cとを備え、診断部162は、診断対象データ取得手段162aと、異常度算出手段162bと、診断手段162cと、寄与度算出手段162dとを備えている。 Here, the detailed configuration of the data mining unit 16 will be described with reference to FIG. 3 (refer to FIG. 1 as appropriate).
As shown in FIG. 3, the data mining unit 16 of this embodiment includes a learning unit 161 and a diagnosis unit 162. The learning unit 161 includes a learning data acquisition unit 161a, a mode division unit 161b, and a model creation unit 161c. The diagnosis unit 162 includes a diagnosis target data acquisition unit 162a, an abnormality degree calculation unit 162b, and a diagnosis unit. 162c and contribution calculation means 162d.
図4は、3つのセンサデータ(センサ1〜センサ3)を例として、センサデータの値の変化の様子と、機械設備2の運転状態(モード)との関係を示したものである。 Here, the mode division will be described with reference to FIG. 4 (refer to FIG. 3 as appropriate).
FIG. 4 shows the relationship between the change in the value of the sensor data and the operation state (mode) of the mechanical equipment 2 using three sensor data (sensor 1 to sensor 3) as an example.
モデル作成手段161cは、モデルデータ記憶手段17に記憶されているモード分割された学習データを用いて、モードごとに事例モデルを作成し、作成した事例モデルのデータであるコードブックをモデルデータ記憶手段17に記憶する。 Returning to FIG. 3, the description of the configuration of the data mining unit 16 will be continued.
The model creation means 161c creates a case model for each mode using the mode-divided learning data stored in the model data storage means 17, and stores a code book which is data of the created case model as model data storage means 17 stored.
診断部162は、モデルデータ記憶手段17に記憶されている事象モデルの学習結果であるコードブックを用いて、センサデータ記憶手段12に記憶されている診断対象となるセンサデータである診断対象データについて、異常度に基づいて異常予兆の有無を診断するものである。また、診断部162は、異常度に対する各センサデータの寄与度を算出する。診断部162は、異常予兆の有無、異常度、寄与度を含む診断結果を診断結果記憶手段18に、機械設備2に対応付けて、診断対象となるセンサデータの測定時刻とともに記憶する。 Returning to FIG. 3, the description of the configuration of the data mining unit 16 will be continued.
The diagnosis unit 162 uses the code book that is the learning result of the event model stored in the model data storage unit 17 to perform diagnosis target data that is sensor data that is a diagnosis target stored in the sensor data storage unit 12. The presence / absence of an abnormality sign is diagnosed based on the degree of abnormality. Further, the diagnosis unit 162 calculates the contribution degree of each sensor data to the degree of abnormality. The diagnosis unit 162 stores a diagnosis result including the presence / absence of an abnormality sign, an abnormality degree, and a contribution degree in the diagnosis result storage unit 18 in association with the mechanical equipment 2 together with the measurement time of sensor data to be diagnosed.
異常度算出手段162bは、順次に診断対象データ取得手段162aから入力される個々の診断対象データについて、コードブックに含まれる複数のクラスタの何れに属するかを判定し、診断対象データが属するクラスタに対応するコードを用いて異常度を算出する。なお、診断対象データは、すべてのクラスタの中で、診断対象データとの距離が最も小さいクラスタに属するものと判定する。 Here, a method for calculating the degree of abnormality will be described with reference to FIGS. 5 and 6 (see FIG. 3 as appropriate).
The degree-of-abnormality calculation means 162b determines which of the plurality of clusters included in the codebook belongs to each diagnosis object data sequentially input from the diagnosis object data acquisition means 162a, and assigns to the cluster to which the diagnosis object data belongs. The degree of abnormality is calculated using the corresponding code. The diagnosis target data is determined to belong to the cluster having the smallest distance from the diagnosis target data among all the clusters.
異常度 = (誤差距離d)/(クラスタ半径r) ・・・ 式(1)
異常度は、クラスタの代表値μで規定されるクラスタ重心から距離が大きいほど、すなわち、クラスタから乖離しているほど大きな値となる。 Returning to FIG. 5, when it is determined that the diagnosis target data p belongs to the cluster C, the distance between the point (cluster centroid) defined by the representative value μ of the cluster C and the point defined by the diagnosis target data p. Is an error distance d, and the degree of abnormality is calculated by the equation (1).
Abnormality = (error distance d) / (cluster radius r) Expression (1)
The degree of abnormality increases as the distance from the cluster center of gravity defined by the cluster representative value μ increases, that is, as the distance from the cluster increases.
診断手段162cは、異常度算出手段162bから入力した異常度に基づいて、異常予兆の有無を診断するものである。診断手段162cは、診断した異常予兆の有無を診断結果の一部として、機械設備2に対応付けて診断結果記憶手段18に記憶する。 Returning to FIG. 3, the description of the configuration of the data mining unit 16 will be continued.
The diagnosis unit 162c diagnoses the presence / absence of an abnormality sign based on the abnormality level input from the abnormality level calculation unit 162b. The diagnosis unit 162c stores, in the diagnosis result storage unit 18, the presence or absence of the diagnosed abnormality sign as a part of the diagnosis result in association with the machine facility 2.
寄与度 = (センサ成分誤差)/(誤差距離d) ・・・ 式(2) Here, the degree of contribution of each sensor data is calculated by Expression (2) using a sensor component error that is an error between the sensor data and a component corresponding to the sensor data of the cluster representative value μ.
Contribution = (sensor component error) / (error distance d) (2)
モデルデータ記憶手段17としては、磁気ディスク装置、光ディスク装置、半導体記憶装置などを用いることができる。 The model data storage means 17 is a storage means for storing the learning data divided for each mode by the mode dividing means 161b and the code book created by the model creating means 161c. The learning data stored in the model data storage unit 17 is referred to by the model creation unit 161c, and the code book is referred to by the abnormality degree calculation unit 162b and the contribution degree calculation unit 162d.
As the model data storage means 17, a magnetic disk device, an optical disk device, a semiconductor storage device, or the like can be used.
図7(a)に示すように、学習データは、個々の機械設備2を識別する識別情報である機械設備2が設置されたサイト(場所)およびそのサイトにおける機械No.(「サイト/機械No.」)に対応付けられて、機械設備2の運転状態を示す「モード」ごとの、各センサ(1〜N)のデータ(「センサ1」〜「センサN」)および「測定時刻」から構成されている。 Here, with reference to FIG. 7 (refer to FIG. 1 as appropriate), the data structure of the learning data and code book stored in the model data storage means 17 will be described.
As shown in FIG. 7A, the learning data includes the site (location) where the machine facility 2 is identification information for identifying the individual machine facility 2, and the machine No. at that site. ("Site / Machine No."), the data ("Sensor 1" to "Sensor N") of each sensor (1-N) for each "mode" indicating the operating state of the mechanical equipment 2 and It consists of “measurement time”.
診断結果記憶手段18としては、磁気ディスク装置、光ディスク装置、半導体記憶装置などを用いることができる。 Returning to FIG. 1, the diagnosis result storage means 18 is a storage means for storing a diagnosis result including the presence / absence of an abnormality sign, an abnormality degree, and a contribution degree by the remote monitoring unit 15 and the data mining unit 16. The diagnosis result stored in the diagnosis result storage unit 18 is referred to by the display control unit 19 and the remote monitoring unit 15.
As the diagnosis result storage means 18, a magnetic disk device, an optical disk device, a semiconductor storage device, or the like can be used.
図8に示すように、診断結果記憶手段18に記憶されるデータマイニング部16による診断結果は、個々の機械設備2を識別する識別情報である機械設備2が設置されたサイト(場所)およびそのサイトにおける機械No.(「サイト/機械No.」)に対応付けられて、「測定時刻」、データマイニング部16で学習データとして用いたセンサデータの測定日を示す「学習データ取得日」、診断対象データが属する「クラスタ番号」、異常予兆の有無を示す「異常フラグ」、機械設備2の運転状態を示す「モード」、「異常度」、各センサ(「センサ1」〜「センサN」)の「測定値」および「学習値」、および各センサ(「センサ1」〜「センサN」)の「寄与度」から構成されている。 Here, referring to FIG. 8 (refer to FIG. 1 as appropriate), the data structure of the diagnosis result by the data mining unit 16 among the diagnosis results stored in the diagnosis result storage means 18 will be described.
As shown in FIG. 8, the diagnosis result by the data mining unit 16 stored in the diagnosis result storage means 18 is a site (location) where the machine equipment 2 which is identification information for identifying each machine equipment 2 is installed, and its site Machine No. at the site (“Site / machine No.”), “measurement time”, “learning data acquisition date” indicating the measurement date of sensor data used as learning data in the data mining unit 16, and “diagnosis target data” “Cluster number”, “abnormal flag” indicating presence / absence of abnormality sign, “mode” indicating operation state of mechanical equipment 2, “abnormality”, “measured value” of each sensor (“sensor 1” to “sensor N”) And “learning value” and “contribution” of each sensor (“sensor 1” to “sensor N”).
また、図8に示した例では、「学習データ取得日」は、クラスタごとに異なる場合がある。この例では、学習データとして、1〜2週間程度の期間のセンサデータを用いて作成する際に、前記したように、この期間のデータを1日ごとに分割し、1日分の学習データを用いて事例モデルをそれぞれ作成して、作成した複数日分の事例モデルの含まれるクラスタの中から、診断対象データが属するクラスタを判定して用いるようにした。「学習データ取得日」とは、診断対象データが属するクラスタの作成に学習データとして用いたセンサデータの測定日を示すものである。 The “learning value” is each sensor component of the representative value μ (see FIG. 4) of the cluster that is determined to belong to the diagnosis target data.
In the example illustrated in FIG. 8, the “learning data acquisition date” may be different for each cluster. In this example, when creating the learning data using the sensor data for a period of about 1 to 2 weeks, as described above, the data for this period is divided every day, and the learning data for one day is obtained. Each case model was created using the cluster, and the cluster to which the diagnosis target data belongs was determined and used from among the clusters containing the created case models for multiple days. The “learning data acquisition date” indicates the measurement date of sensor data used as learning data for creating a cluster to which diagnosis target data belongs.
図9は、上段および中段に、センサデータ選択メニューB2で選択された2個のセンサデータについて、測定値(実線)、学習値(破線)および寄与度(点線)の変化の履歴をグラフ表示し、下段に異常度の変化の履歴をグラフ表示したものである。 Here, a display example of the diagnosis result will be described with reference to FIGS. 9 and 10.
FIG. 9 is a graph showing the change history of the measured value (solid line), the learned value (dashed line), and the contribution (dotted line) for the two sensor data selected in the sensor data selection menu B2 in the upper and middle stages. In the lower part, the change history of the degree of abnormality is displayed as a graph.
次に、図11を参照(適宜図1ないし図3参照)して、本実施形態における異常予兆診断装置1による異常予兆診断の動作について説明する。
なお、ここでは、所定期間以上に渡って、機械設備2に設置された複数のセンサによって測定されたセンサデータおよび機械設備2の保守情報が、それぞれセンサデータ記憶手段12および保守情報記憶手段14に蓄積されているものとする。
また、この異常予兆診断処理を行うタイミングは、診断対象となる機械設備2の性質や運用状況に応じて定められるべきものである。例えば、発電プラントなどの機械設備2を診断するためには、この処理は、例えば、1日に1回、定められた時刻に行われるものである。 [Operation of the abnormal sign diagnosis device]
Next, referring to FIG. 11 (refer to FIGS. 1 to 3 as appropriate), the operation of the abnormal sign diagnosis by the abnormal sign diagnosis apparatus 1 in the present embodiment will be described.
Here, sensor data measured by a plurality of sensors installed in the mechanical equipment 2 and maintenance information of the mechanical equipment 2 are stored in the sensor data storage means 12 and the maintenance information storage means 14 respectively for a predetermined period or more. It is assumed that it has been accumulated.
In addition, the timing for performing the abnormality sign diagnosis process should be determined according to the nature and operation status of the machine equipment 2 to be diagnosed. For example, in order to diagnose the mechanical equipment 2 such as a power plant, this process is performed, for example, once a day at a predetermined time.
なお、データマイニング部16によって異常予兆があると診断した場合に、診断すべき異常種別に係る総合診断手段1521〜1522における診断ルールを再構築するものとする。ここでは、診断すべき異常種別に対応する総合診断手段が、総合診断手段1521であるとして以下の説明をする。 Next, the abnormality sign diagnosis apparatus 1 uses the individual diagnosis means 151 11 to 151 1N and the individual diagnosis means 151 21 to 151 2N based on the contribution calculated in step S14 by the comprehensive diagnosis means 152 1 to 152 2 . The diagnosis rule for the comprehensive diagnosis of the abnormal sign using the diagnosis result is corrected (step S15).
In addition, when the data mining unit 16 diagnoses that there is an abnormality sign, the diagnosis rules in the comprehensive diagnosis units 152 1 to 152 2 relating to the abnormality type to be diagnosed are reconstructed. Here, overall diagnosis means corresponding to the abnormal kind to be diagnosed, the following described as a comprehensive diagnostic means 152 1.
図12に示すように、異常予兆診断装置1は、まず、学習データ取得手段161aによって、センサデータ記憶手段12から、予め定められた所定期間に測定されたセンサデータを学習データとして取得する(ステップS20)。 Next, with reference to FIG. 12 (refer to FIG. 1 and FIG. 3 as appropriate), an example model creation process will be described. This process corresponds to the process of step S10 in FIG.
As shown in FIG. 12, the abnormality sign diagnosis apparatus 1 first acquires, as learning data, sensor data measured in a predetermined period from the sensor data storage unit 12 by the learning data acquisition unit 161a (Step S1). S20).
2 機械設備
10 通信手段
11 センサデータ取得手段
12 センサデータ記憶手段
13 保守情報取得手段
14 保守情報記憶手段
15 リモートモニタリング部(第2診断手段)
16 データマイニング部(第1診断手段)
17 モデルデータ記憶手段
18 診断結果記憶手段
19 表示制御手段
20 表示手段
15111〜1512N 個別診断手段
1521〜1522 総合診断手段
161 学習部
161a 学習データ取得手段
161b モード分割手段
161c モデル作成手段
162 診断部
162a 診断対象データ取得手段
162b 異常度算出手段
162c 診断手段
162d 寄与度算出手段 DESCRIPTION OF SYMBOLS 1 Abnormality sign diagnostic apparatus 2 Mechanical equipment 10 Communication means 11 Sensor data acquisition means 12 Sensor data storage means 13 Maintenance information acquisition means 14 Maintenance information storage means 15 Remote monitoring part (2nd diagnosis means)
16 Data mining department (first diagnostic means)
17 Model Data Storage Unit 18 Diagnosis Result Storage Unit 19 Display Control Unit 20 Display Unit 151 11 to 151 2N Individual Diagnosis Unit 152 1 to 152 2 Total Diagnosis Unit 161 Learning Unit 161a Learning Data Acquisition Unit 161b Mode Division Unit 161c Model Creation Unit 162 Diagnosis unit 162a Diagnosis target data acquisition unit 162b Abnormality calculation unit 162c Diagnosis unit 162d Contribution calculation unit
Claims (10)
- 機械設備の異常予兆の有無を診断する異常予兆診断装置であって、
前記機械設備に設置された複数のセンサによって測定された多次元センサデータを取得するセンサデータ取得手段と、
異常予兆の診断対象となる前記多次元センサデータである診断対象データについて、前記機械設備が正常に稼動しているときに取得した前記多次元センサデータを学習データとして用い、当該学習データをクラスタ化して作成された事例モデルからの前記診断対象データの乖離の度合いを示す異常度の大きさに基づいて、前記異常予兆の有無を診断する第1診断手段と、
前記診断対象データを構成する1または2以上の個別のセンサデータの値が、それぞれ予め定められた所定の範囲内にあるか否かに基づいて、前記異常予兆の有無を診断する第2診断手段と、を備え、
前記第1診断手段によって異常予兆があると診断された場合に、前記第2診断手段が異常予兆診断のために参照する前記1または2以上の個別のセンサデータを、前記異常度に対する前記個別のセンサデータの寄与の大きさを示す寄与度の大きさに基づいて、前記診断対象データの中から選択することを特徴とする異常予兆診断装置。 An abnormality sign diagnosis device for diagnosing the presence or absence of an abnormality sign of mechanical equipment,
Sensor data acquisition means for acquiring multidimensional sensor data measured by a plurality of sensors installed in the mechanical facility;
For the diagnosis target data, which is the multi-dimensional sensor data to be diagnosed for an abnormality sign, the multi-dimensional sensor data acquired when the mechanical equipment is operating normally is used as learning data, and the learning data is clustered. First diagnosis means for diagnosing the presence or absence of the abnormality sign based on the magnitude of the degree of abnormality indicating the degree of deviation of the diagnosis target data from the case model created
Second diagnostic means for diagnosing the presence / absence of the abnormality sign based on whether or not the value of one or more individual sensor data constituting the diagnosis target data is within a predetermined range. And comprising
When the first diagnostic means diagnoses that there is an abnormality sign, the one or more individual sensor data referred to by the second diagnosis means for abnormality sign diagnosis is the individual sensor data for the degree of abnormality. An abnormality predictor diagnosis apparatus, wherein the diagnosis target data is selected from the diagnosis target data based on a degree of contribution indicating a magnitude of contribution of sensor data. - 前記第2診断手段が異常予兆診断のために参照するセンサデータは、前記寄与度が最も大きいセンサデータであることを特徴とする請求項1に記載の異常予兆診断装置。 2. The abnormality sign diagnosis apparatus according to claim 1, wherein sensor data referred to by the second diagnosis unit for abnormality sign diagnosis is sensor data having the largest contribution.
- 前記第1診断手段によって異常予兆があると診断された場合に、前記第2診断手段が診断のために参照したセンサデータの変化の履歴をグラフ表示することを特徴とする請求項1または請求項2に記載の異常予兆診断装置。 The history of changes in sensor data referred to by the second diagnostic unit for diagnosis is displayed in a graph when the first diagnostic unit diagnoses that there is a sign of abnormality. 2. The abnormal sign diagnostic apparatus according to 2.
- 前記異常度は、前記診断対象データと、前記事例モデルを構成するクラスタの中で前記診断対象データとクラスタの重心との距離が最も近いクラスタである所属クラスタの重心と、の距離を、当該所属クラスタの広がりを示す指標であるクラスタ半径で除した値であることを特徴とする請求項1ないし請求項3の何れか一項に記載の異常予兆診断装置。 The degree of abnormality is the distance between the diagnosis target data and the centroid of the affiliation cluster that is the closest distance between the diagnosis target data and the centroid of the cluster among the clusters constituting the case model. The abnormality sign diagnosis apparatus according to any one of claims 1 to 3, wherein the abnormality sign diagnosis apparatus is a value divided by a cluster radius that is an index indicating a cluster spread.
- 前記寄与度は、前記診断対象データを構成する個別のセンサデータと、前記所属クラスタの重心の当該個別のセンサデータに対応する成分との差の絶対値を、前記診断対象データと前記所属クラスタの重心との距離で除した値であることを特徴とする請求項1ないし請求項4の何れか一項に記載の異常予兆診断装置。 The contribution degree is obtained by calculating an absolute value of a difference between individual sensor data constituting the diagnosis target data and a component corresponding to the individual sensor data of the center of gravity of the belonging cluster, and calculating the contribution of the diagnosis target data and the belonging cluster. The abnormality sign diagnosis apparatus according to any one of claims 1 to 4, wherein the abnormality sign diagnosis apparatus is a value divided by a distance from a center of gravity.
- 機械設備の異常予兆の有無を診断する異常予兆診断方法であって、
前記機械設備に設置された複数のセンサによって測定された多次元センサデータを取得するセンサデータ取得工程と、
異常予兆の診断対象となる前記多次元センサデータである診断対象データについて、前記機械設備が正常に稼動しているときに取得した前記多次元センサデータを学習データとして用い、当該学習データをクラスタ化して作成された事例モデルからの乖離の度合いを示す異常度の大きさに基づいて、前記異常予兆の有無を診断する第1診断工程と、
前記診断対象データを構成する1または2以上の個別のセンサデータの値が、それぞれ予め定められた所定の範囲内にあるか否かに基づいて、前記異常予兆の有無を診断する第2診断工程と、を含み、
前記第1診断工程において異常予兆があると診断された場合に、前記第2診断工程における異常予兆診断のために参照する前記1または2以上の個別のセンサデータを、前記異常度に対する前記個別のセンサデータの寄与の大きさを示す寄与度の大きさに基づいて、前記診断対象データの中から選択することを特徴とする異常予兆診断方法。 An abnormality sign diagnosis method for diagnosing the presence or absence of an abnormality sign of mechanical equipment,
A sensor data acquisition step of acquiring multidimensional sensor data measured by a plurality of sensors installed in the mechanical facility;
For the diagnosis target data, which is the multi-dimensional sensor data to be diagnosed for an abnormality sign, the multi-dimensional sensor data acquired when the mechanical equipment is operating normally is used as learning data, and the learning data is clustered. A first diagnosis step of diagnosing the presence or absence of the abnormality sign based on the magnitude of the degree of abnormality indicating the degree of deviation from the created case model;
A second diagnosis step of diagnosing the presence / absence of the abnormality sign based on whether or not the value of one or more individual sensor data constituting the diagnosis object data is within a predetermined range, respectively. And including
When it is diagnosed that there is an abnormality sign in the first diagnosis step, the one or more individual sensor data to be referred to for abnormality sign diagnosis in the second diagnosis step An abnormality sign diagnosis method, wherein the diagnosis target data is selected based on a degree of contribution indicating a magnitude of contribution of sensor data. - 前記第2診断工程において異常予兆診断のために参照するセンサデータは、前記寄与度が最も大きいセンサデータであることを特徴とする請求項6に記載の異常予兆診断方法。 The abnormality sign diagnosis method according to claim 6, wherein sensor data referred to for abnormality sign diagnosis in the second diagnosis step is sensor data having the largest contribution.
- 前記第1診断工程において異常予兆があると診断された場合に、前記第2診断工程で診断のために参照したセンサデータの変化の履歴をグラフ表示することを特徴とする請求項6または請求項7に記載の異常予兆診断方法。 The history of changes in sensor data referred to for diagnosis in the second diagnosis step is displayed in a graph when it is diagnosed that there is an abnormality sign in the first diagnosis step. 8. The abnormal sign diagnostic method according to 7.
- 前記異常度は、前記診断対象データと、前記事例モデルを構成するクラスタの中で前記診断対象データとクラスタの重心との距離が最も近いクラスタである所属クラスタの重心と、の距離を、当該所属クラスタの広がりを示す指標であるクラスタ半径で除した値であることを特徴とする請求項6ないし請求項8の何れか一項に記載の異常予兆診断方法。 The degree of abnormality is the distance between the diagnosis target data and the centroid of the affiliation cluster that is the closest distance between the diagnosis target data and the centroid of the cluster among the clusters constituting the case model. The abnormality sign diagnosis method according to any one of claims 6 to 8, wherein the abnormality sign diagnosis method is a value divided by a cluster radius which is an index indicating a cluster spread.
- 前記寄与度は、前記診断対象データを構成する個別のセンサデータと、前記所属クラスタの重心の当該個別のセンサデータに対応する成分との差の絶対値を、前記診断対象データと前記所属クラスタの重心との距離で除した値であることを特徴とする請求項6ないし請求項9の何れか一項に記載の異常予兆診断方法。 The contribution degree is obtained by calculating an absolute value of a difference between individual sensor data constituting the diagnosis target data and a component corresponding to the individual sensor data of the center of gravity of the belonging cluster, and calculating the contribution of the diagnosis target data and the belonging cluster. The abnormality sign diagnosis method according to claim 6, wherein the abnormality sign diagnosis method is a value divided by a distance from the center of gravity.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011138941A JP4832609B1 (en) | 2011-06-22 | 2011-06-22 | Abnormal sign diagnosis device and abnormality sign diagnosis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011138941A JP4832609B1 (en) | 2011-06-22 | 2011-06-22 | Abnormal sign diagnosis device and abnormality sign diagnosis method |
Publications (2)
Publication Number | Publication Date |
---|---|
JP4832609B1 true JP4832609B1 (en) | 2011-12-07 |
JP2013008111A JP2013008111A (en) | 2013-01-10 |
Family
ID=45418171
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
JP2011138941A Active JP4832609B1 (en) | 2011-06-22 | 2011-06-22 | Abnormal sign diagnosis device and abnormality sign diagnosis method |
Country Status (1)
Country | Link |
---|---|
JP (1) | JP4832609B1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103198147A (en) * | 2013-04-19 | 2013-07-10 | 上海岩土工程勘察设计研究院有限公司 | Method for distinguishing and processing abnormal automatized monitoring data |
JP2013175108A (en) * | 2012-02-27 | 2013-09-05 | Mitsubishi Electric Corp | Clustering device and clustering program |
JP2013214292A (en) * | 2012-03-06 | 2013-10-17 | Jfe Steel Corp | Abnormality monitoring system and abnormality monitoring method |
JP6466617B1 (en) * | 2018-07-17 | 2019-02-06 | 株式会社日立パワーソリューションズ | Abnormal sign diagnosis system and abnormality sign diagnosis method |
JP2020061054A (en) * | 2018-10-12 | 2020-04-16 | 株式会社日立パワーソリューションズ | Environmental factor setting system |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6146275B2 (en) * | 2013-11-27 | 2017-06-14 | 株式会社豊田中央研究所 | Fault diagnosis support system, fault diagnosis support apparatus, and fault diagnosis support method |
JP5684941B1 (en) * | 2014-07-31 | 2015-03-18 | 株式会社日立パワーソリューションズ | Abnormal sign diagnostic apparatus and abnormal sign diagnostic method |
JP6482817B2 (en) * | 2014-10-22 | 2019-03-13 | 株式会社日立製作所 | Plant monitoring support system and plant monitoring support method |
US10719577B2 (en) | 2014-12-05 | 2020-07-21 | Nec Corporation | System analyzing device, system analyzing method and storage medium |
JP6297202B2 (en) * | 2015-02-19 | 2018-03-20 | 三菱電機株式会社 | Procedure extraction system |
JP6328071B2 (en) * | 2015-03-31 | 2018-05-23 | 東芝メモリ株式会社 | Abnormal sign detection system and semiconductor device manufacturing method |
JP5875726B1 (en) | 2015-06-22 | 2016-03-02 | 株式会社日立パワーソリューションズ | Preprocessor for abnormality sign diagnosis apparatus and processing method thereof |
JP6599727B2 (en) * | 2015-10-26 | 2019-10-30 | 株式会社Screenホールディングス | Time-series data processing method, time-series data processing program, and time-series data processing apparatus |
JP6450858B2 (en) * | 2015-11-25 | 2019-01-09 | 株式会社日立製作所 | Equipment management apparatus and method |
JP6467364B2 (en) * | 2016-02-24 | 2019-02-13 | 日本電信電話株式会社 | Monitoring control system and rule generation method |
JP6686593B2 (en) | 2016-03-23 | 2020-04-22 | 日本電気株式会社 | Data processing device, data processing system, data processing method and program |
JP2018097662A (en) | 2016-12-14 | 2018-06-21 | オムロン株式会社 | Control device, control program and control method |
EP3557354A4 (en) | 2016-12-14 | 2020-03-04 | Omron Corporation | Control device, control program, and control method |
JP2018156151A (en) * | 2017-03-15 | 2018-10-04 | ファナック株式会社 | Abnormality detecting apparatus and machine learning device |
JP2018159981A (en) | 2017-03-22 | 2018-10-11 | オムロン株式会社 | Control system, control apparatus, control program, and control method |
US20200089734A1 (en) * | 2017-05-08 | 2020-03-19 | Hitachi, Ltd. | Time series data analysis control method and analysis control device |
JP6721563B2 (en) | 2017-11-28 | 2020-07-15 | ファナック株式会社 | Numerical control device |
JP2019196036A (en) | 2018-05-07 | 2019-11-14 | トヨタ自動車株式会社 | Diagnostic apparatus, diagnostic system, and diagnostic method |
JP2020046783A (en) * | 2018-09-14 | 2020-03-26 | 株式会社東芝 | Process-monitoring support device, process-monitoring support system, process-monitoring support method, process-monitoring support program, and terminal device |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011070635A (en) * | 2009-08-28 | 2011-04-07 | Hitachi Ltd | Method and device for monitoring state of facility |
JP2011081697A (en) * | 2009-10-09 | 2011-04-21 | Hitachi Ltd | Facility condition monitoring method, monitoring system, and monitoring program |
-
2011
- 2011-06-22 JP JP2011138941A patent/JP4832609B1/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011070635A (en) * | 2009-08-28 | 2011-04-07 | Hitachi Ltd | Method and device for monitoring state of facility |
JP2011081697A (en) * | 2009-10-09 | 2011-04-21 | Hitachi Ltd | Facility condition monitoring method, monitoring system, and monitoring program |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013175108A (en) * | 2012-02-27 | 2013-09-05 | Mitsubishi Electric Corp | Clustering device and clustering program |
JP2013214292A (en) * | 2012-03-06 | 2013-10-17 | Jfe Steel Corp | Abnormality monitoring system and abnormality monitoring method |
CN103198147A (en) * | 2013-04-19 | 2013-07-10 | 上海岩土工程勘察设计研究院有限公司 | Method for distinguishing and processing abnormal automatized monitoring data |
CN103198147B (en) * | 2013-04-19 | 2016-05-11 | 上海顺凯信息技术股份有限公司 | The differentiation of automatic monitoring abnormal data and processing method |
JP6466617B1 (en) * | 2018-07-17 | 2019-02-06 | 株式会社日立パワーソリューションズ | Abnormal sign diagnosis system and abnormality sign diagnosis method |
JP2020012691A (en) * | 2018-07-17 | 2020-01-23 | 株式会社日立パワーソリューションズ | Abnormality sign diagnosing system and abnormality sign diagnosing method |
JP2020061054A (en) * | 2018-10-12 | 2020-04-16 | 株式会社日立パワーソリューションズ | Environmental factor setting system |
Also Published As
Publication number | Publication date |
---|---|
JP2013008111A (en) | 2013-01-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10545487B2 (en) | Interactive diagnostic system and method for managing process model analysis | |
US9864665B2 (en) | Adaptive handling of operating data based on assets' external conditions | |
US10527520B2 (en) | Operating wind motors and determining their remaining useful life | |
Wilkinson et al. | Comparison of methods for wind turbine condition monitoring with SCADA data | |
US10018007B2 (en) | Systems and methods to visualize component health and preventive maintenance needs for subsea control subsystem components | |
Niu | Data-driven technology for engineering systems health management | |
US5210704A (en) | System for prognosis and diagnostics of failure and wearout monitoring and for prediction of life expectancy of helicopter gearboxes and other rotating equipment | |
US9261862B2 (en) | Automation management system and method | |
JP4850857B2 (en) | Adaptive modeling of change states in predictive state monitoring | |
US7088255B2 (en) | Health monitoring display system for a complex plant | |
JP5306902B2 (en) | System and method for high performance condition monitoring of asset systems | |
JP2014059910A (en) | Method for monitoring state of facility | |
JP5297272B2 (en) | Device abnormality monitoring method and system | |
CN106873548B (en) | Plant process management system with standard asset health | |
US20180365089A1 (en) | Abnormality detection system, abnormality detection method, abnormality detection program, and method for generating learned model | |
US6694286B2 (en) | Method and system for monitoring the condition of an individual machine | |
JP5284503B2 (en) | Diagnostic system and method for predictive condition monitoring | |
US20150220847A1 (en) | Information Processing Apparatus, Diagnosis Method, and Program | |
EP2477086A1 (en) | Anomaly detection and diagnostic method, anomaly detection and diagnostic system, and anomaly detection and diagnostic program | |
US20030216888A1 (en) | Predictive maintenance display system | |
WO2013111397A1 (en) | Plant or facility fault determination method and health management method | |
JP5530020B1 (en) | Abnormality diagnosis system and abnormality diagnosis method | |
JP5179086B2 (en) | Industrial process monitoring method and monitoring system | |
US8803702B2 (en) | Instrument status displaying device and instrument status displaying method | |
JP5684941B1 (en) | Abnormal sign diagnostic apparatus and abnormal sign diagnostic method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
TRDD | Decision of grant or rejection written | ||
A01 | Written decision to grant a patent or to grant a registration (utility model) |
Free format text: JAPANESE INTERMEDIATE CODE: A01 |
|
A61 | First payment of annual fees (during grant procedure) |
Free format text: JAPANESE INTERMEDIATE CODE: A61 Effective date: 20110920 |
|
R150 | Certificate of patent or registration of utility model |
Ref document number: 4832609 Country of ref document: JP Free format text: JAPANESE INTERMEDIATE CODE: R150 Free format text: JAPANESE INTERMEDIATE CODE: R150 |
|
FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20140930 Year of fee payment: 3 |
|
S533 | Written request for registration of change of name |
Free format text: JAPANESE INTERMEDIATE CODE: R313533 |
|
R350 | Written notification of registration of transfer |
Free format text: JAPANESE INTERMEDIATE CODE: R350 |