TW202125138A - Abnormality diagnosis device and program capable of accurately determining a state of a device to be diagnosed - Google Patents
Abnormality diagnosis device and program capable of accurately determining a state of a device to be diagnosed Download PDFInfo
- Publication number
- TW202125138A TW202125138A TW109145508A TW109145508A TW202125138A TW 202125138 A TW202125138 A TW 202125138A TW 109145508 A TW109145508 A TW 109145508A TW 109145508 A TW109145508 A TW 109145508A TW 202125138 A TW202125138 A TW 202125138A
- Authority
- TW
- Taiwan
- Prior art keywords
- data
- measurement
- failure mode
- failure
- abnormality
- Prior art date
Links
Images
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
Description
本發明係關於一種異常診斷裝置及程式。The present invention relates to an abnormal diagnosis device and program.
於監視風車等機械之狀態時,大多情況下使用機械所裝備之感測器等,以固定頻率收集機械之溫度或加速度等物理量,並對收集到之資料進行分析處理,藉此,監視機械之狀態,進行正常/異常之判斷。近年來,提案有一種活用人工智慧技術之一領域即機械學習技術,學習過去收集之機械之運轉資料(亦稱為“訓練”),且推定、判斷當前之狀態的異常診斷方法(下述專利文獻1及非專利文獻1)。例如,下述專利文獻1之要約書中,有如下記載:「設備狀態監視裝置具備:異常度算出模型製成部,其基於自監視對象設備正常時之狀態量變動資料擷取之正常時特徵量群,製成用以算出監視對象設備之監視時之監視時特徵量群之異常度的異常度算出模型;異常度算出部,其使用異常度算出模型,算出監視時特徵量群之異常度;異常判定部,其基於異常度,判定監視對象設備有無異常;異常貢獻度算出部,其求出構成算出由異常判定部判定有異常之異常度所用之監視時特徵量群的複數個特徵量各者對異常度之貢獻度;及異常原因特定部,其基於表示貢獻度及監視對象設備之異常原因與複數個特徵量之關係性的因果矩陣,特定異常原因」。
[先前技術文獻]
[專利文獻]When monitoring the status of windmills and other machinery, in most cases, sensors equipped with the machinery are used to collect physical quantities such as the temperature or acceleration of the machinery at a fixed frequency, and the collected data are analyzed and processed to monitor the machinery’s performance. Status, judge normal/abnormal. In recent years, there has been a proposal for an abnormal diagnosis method that utilizes one of the fields of artificial intelligence technology, namely machine learning technology, learns the operating data of the machine collected in the past (also called "training"), and estimates and judges the current state (the following
[專利文獻1]日本專利特開2019-128704號公報 [非專利文獻][Patent Document 1] Japanese Patent Laid-Open No. 2019-128704 [Non-Patent Literature]
[非專利文獻1]鈴木、內山、湯田:利用資料探勘之異常檢測技術。運籌學,2012年9月號,pp.506~511[Non-Patent Document 1] Suzuki, Uchiyama, and Yuda: Anomaly detection technology using data exploration. Operations Research, September 2012, pp.506~511
[發明所欲解決之問題]
於上述之機械學習之異常診斷方法中,將來自監視對象設備之測定資料之一部分作為輸入資料使用。以下,將使用於異常診斷方法之輸入之測定資料稱為“特徵量”或“特徵量群”。風力發電裝置等構造複雜之裝置中,測定資料之種類與數量龐大。因此,根據作為特徵量群應用之測定資料之選擇狀態,有異常診斷之結果大幅變化之情形。先前,以使用者之主觀判斷選擇特徵量群之情形較多,而有異常診斷之結果不正確之情形。
如上所述,專利文獻1中,記載有「算出監視時特徵量群之異常度」、「求出複數個特徵量之各者對異常度之貢獻度」,但如此一來,若「用於異常診斷之特徵量群」之選擇不合適,則有「異常度」或「貢獻度」之算出結果不合適之問題。
本發明係鑑於上述之事項而完成者,目的在於提供一種可正確判定診斷對象裝置之狀態之異常診斷裝置及程式。
[解決問題之技術手段][The problem to be solved by the invention]
In the above-mentioned abnormal diagnosis method of machine learning, a part of the measurement data from the monitoring target device is used as the input data. Hereinafter, the measurement data used for the input of the abnormality diagnosis method is referred to as "feature quantity" or "feature quantity group". In devices with complex structures such as wind power generation devices, the types and quantities of measurement data are huge. Therefore, depending on the selection status of the measurement data used as the feature quantity group, the result of the abnormality diagnosis may vary greatly. Previously, there were many cases in which the feature quantity group was selected based on the subjective judgment of the user, and there were cases in which the result of the abnormal diagnosis was incorrect.
As mentioned above,
用以解決上述問題之本發明之異常診斷裝置之特徵在於具備:資料庫,其記憶診斷對象裝置中之複數個測定項目之測定資料;故障模式選擇部,其選擇成為檢測對象之故障模式;示教資料製成部,其基於上述診斷對象裝置或其他裝置中之與上述故障模式對應之資料即故障對應測定資料,製成用以判定上述診斷對象裝置有無故障之示教資料;測定項目重要度算出部,其基於上述示教資料,算出上述故障模式中之上述測定項目之重要度;特徵量選定部,其基於算出之上述重要度,選擇上述測定項目之一部分作為對於上述故障模式之特徵量;異常度算出部,其基於上述特徵量相關之上述測定資料,算出對應於上述故障模式之異常度;及裝置狀態判定部,其基於算出之上述異常度,判定上述診斷對象裝置之狀態。 [發明之效果]The abnormality diagnosis device of the present invention for solving the above-mentioned problems is characterized by having: a database that stores measurement data of a plurality of measurement items in the diagnosis target device; a failure mode selection unit that selects the failure mode to be the detection target; The teaching data preparation unit, which generates teaching data for judging whether the diagnosis target device has a failure based on the data corresponding to the failure mode in the diagnosis target device or other devices, that is, the failure correspondence measurement data; the importance of the measurement item A calculation unit that calculates the importance of the measurement item in the failure mode based on the teaching data; a feature quantity selection unit that selects a part of the measurement item as the feature quantity for the failure mode based on the calculated importance An abnormality calculation unit that calculates an abnormality corresponding to the failure mode based on the measurement data related to the characteristic amount; and a device state determination unit that determines the state of the diagnosis target device based on the calculated abnormality. [Effects of Invention]
根據本發明,可正確地判定診斷對象裝置之狀態。According to the present invention, the state of the device to be diagnosed can be accurately determined.
[第1實施形態]
<第1實施形態之構成>
圖1係較佳之第1實施形態之異常診斷系統1之方塊圖。
圖1中,異常診斷系統1具備診斷對象裝置10、感測器部20、及異常診斷裝置100(電腦)。另,本實施形態中,診斷對象裝置10為風力發電裝置。感測器部20具備計測診斷對象裝置10之各部之溫度、壓力、加速度等物理量之N個(N為複數個)感測器22-1~22-N。且,感測器部20以特定之取樣週期取樣感測器22-1~22-N之測定結果即物理量,並將該結果作為測定資料DM供給至異常診斷裝置100。異常診斷裝置100基於自感測器部20供給之測定資料DM,對診斷對象裝置10之狀態進行診斷。[First Embodiment]
<Configuration of the first embodiment>
Fig. 1 is a block diagram of an
異常診斷裝置100具備CPU(Central Processing Unit:中央處理單元)、RAM(Random Access Memory:隨機存取記憶體)、ROM(Read Only Memory:唯讀記憶體)、SSD(Solid State Drive:固態驅動器)等作為一般電腦之硬體,SSD中儲存有OS(Operating System:操作系統)、應用程式、各種資料等。OS及應用程式於RAM中展開,且由CPU執行。圖1中,異常診斷裝置100之內部將藉由應用程式等實現之功能顯示為方塊。The
即,異常診斷裝置100具備:測定資料取得部101、故障資料取得部102、示教資料製成部103(示教資料製成機構)、測定項目重要度算出部104(測定項目重要度算出機構)、特徵量選定部105(特徵量選定機構)、故障模式選擇部106(故障模式選擇機構)、異常度算出部107(異常度算出機構)、裝置狀態判定部108(裝置狀態判定機構)及資料庫110(資料庫機構)。That is, the
測定資料取得部101自感測器部20取得測定資料DM。資料庫110具備一個或複數個記憶裝置,且以電子檔案形式儲存各種資料。例如,資料庫110記憶以下所列舉之資料。關於各個資料之內容稍後敘述。 ·測定資料取得部101取得之測定資料DM ·過去之故障資料DF ·示教資料DT ·表示測定資料之重要度之重要度資料DQ、 ·異常診斷之結果即異常度A之時間序列資料、 ·狀態資料DC、 ·故障模式列表MLThe measurement data acquisition unit 101 acquires the measurement data DM from the sensor unit 20. The database 110 has one or more memory devices, and stores various data in the form of electronic files. For example, the database 110 stores the data listed below. The content of each material will be described later. ·Measurement data DM acquired by the measurement data acquisition section 101 ·Past fault data DF ·Teaching data DT ·Importance data DQ, which indicates the importance of measurement data, ·The result of abnormal diagnosis is the time series data of abnormal degree A, ·Status data DC, ·Failure mode list ML
圖2係顯示測定資料DM之資料構造之一例之圖。 圖示之例中,取樣時刻ts_1、ts_2、…ts_max為每個取樣週期之時刻。且,測定資料DM包含各個感測器22-1~22-N(換言之,N個測定項目P1~PN)在取樣時刻ts_1、ts_2、…ts_max之計測結果。如此,測定資料DM為自各感測器22-1~22-N收集之時間序列資料之集合,可表記為如圖2之矩陣。Figure 2 is a diagram showing an example of the data structure of the measurement data DM. In the example shown in the figure, the sampling times ts_1, ts_2, ... ts_max are the times of each sampling period. In addition, the measurement data DM includes the measurement results of the respective sensors 22-1 to 22-N (in other words, N measurement items P1 to PN) at the sampling times ts_1, ts_2, ... ts_max. In this way, the measurement data DM is a collection of time series data collected from the sensors 22-1-22-N, which can be expressed as a matrix as shown in FIG. 2.
返回至圖1,故障資料取得部102於診斷對象裝置10中發現故障時,將與該故障關聯之故障資料DF記錄於資料庫110。
圖3係顯示故障資料DF之一例之圖。
如圖所示,故障資料DF包含故障發現時日DFT、故障模式DFM及故障對應測定資料DFK。此處,故障發現時日DFT為表示使用者或故障監視裝置(未圖示)發現診斷對象裝置10之故障之時日的資料。Returning to FIG. 1, when the fault
又,故障模式DFM為與發現之故障之具體內容對應之資料,例如,如圖所示,舉出「增速機中速軸小齒輪損傷」、「發電機軸承損傷」等。又,故障對應測定資料DFK為上述之測定資料DM之一部分,且等同於至故障發現時日DFT為止之特定期間內之測定資料DM。In addition, the failure mode DFM is data corresponding to the specific content of the discovered failure. For example, as shown in the figure, "damage to the pinion of the speed increaser's shaft", "damage to the generator bearing", etc. are mentioned. In addition, the failure-corresponding measurement data DFK is a part of the above-mentioned measurement data DM, and is equivalent to the measurement data DM within a specific period until the DFT when the failure is discovered.
返回至圖1,記憶於資料庫110之故障模式列表ML為列舉可作為上述之故障模式DFM(參照圖3)選擇之各種故障模式的列表。又,示教資料製成部103基於上述之故障資料DF,產生示教資料DT。Returning to FIG. 1, the failure mode list ML stored in the database 110 is a list of various failure modes that can be selected as the failure mode DFM (refer to FIG. 3) described above. In addition, the teaching
圖4係顯示示教資料DT之一例之圖。 圖示之例中,取樣時刻ts_a、ts_b、ts_c、ts_d等為每個取樣週期之時刻。然而,圖4中,自下而上時刻進展,取樣時刻ts_d等同於故障發現時日DFT(參照圖3)。取樣時刻ts_c為較ts_d往前特定時間T1(第1特定時間)之時刻,取樣時刻ts_b為ts_c之上一個取樣時刻。又,取樣時刻ts_a為較ts_b往前特定時間T2(第2特定時間)之時刻。Fig. 4 is a diagram showing an example of the teaching data DT. In the example shown in the figure, the sampling time ts_a, ts_b, ts_c, ts_d, etc. are the time of each sampling period. However, in FIG. 4, the time progresses from bottom to top, and the sampling time ts_d is equivalent to the DFT when the fault is found (refer to FIG. 3). The sampling time ts_c is a specific time T1 (first specific time) before ts_d, and the sampling time ts_b is the next sampling time before ts_c. In addition, the sampling time ts_a is a specific time T2 (second specific time) before ts_b.
且,示教資料DT包含感測器22-1~22-N(參照圖1)針對測定項目P1~PN之各者在取樣時刻ts_a~ts_d之計測結果。再者,示教資料DT包含異常旗標DTF之旗標。換言之,示教資料DT包含故障資料DF(參照圖3)所含之故障對應測定資料DFK與異常旗標DTF。異常旗標DTF相對於取樣時刻ts_c~ts_d為“1”,相對於取樣時刻ts_a~ts_b為“0”。In addition, the teaching data DT includes the measurement results of the sensors 22-1 to 22-N (see FIG. 1) for each of the measurement items P1 to PN at the sampling times ts_a to ts_d. Furthermore, the teaching data DT includes the flag of the abnormal flag DTF. In other words, the teaching data DT includes the fault corresponding measurement data DFK and the abnormal flag DTF contained in the fault data DF (refer to FIG. 3). The abnormality flag DTF is "1" with respect to the sampling times ts_c to ts_d, and is "0" with respect to the sampling times ts_a to ts_b.
此處,異常旗標DTF為“1”之取樣時刻為「發生異常之可能性高」之時刻,為“0”之取樣時刻為「發生異常之可能性低」之時刻。且,將異常旗標DTF為“1”之範圍之示教資料DT稱為異常時示教資料DT1,將為“0”之範圍之示教資料DT稱為正常時示教資料DT2。此處,特定時間T1、T2之值可根據與對應之故障模式DFM(參照圖3)關聯之零件知識、或診斷對象裝置10之運轉歷程等而決定。Here, the sampling time when the abnormality flag DTF is "1" is the time when the probability of occurrence of abnormality is high, and the sampling time when the abnormality flag DTF is "1" is the time when the probability of occurrence of abnormality is low. In addition, the teaching data DT in the range where the abnormality flag DTF is "1" is called the teaching data DT1 at the time of abnormality, and the teaching data DT in the range of "0" is called the teaching data DT2 in the normal time. Here, the values of the specific times T1 and T2 can be determined based on the part knowledge associated with the corresponding failure mode DFM (refer to FIG. 3), or the operation history of the
返回至圖1,測定項目重要度算出部104基於上述之示教資料DT,算出重要度資料DQ。此處,重要度資料DQ包含分別與N個測定項目(感測器22-1~22-N之測定結果)對應之N個重要度Q1~QN。N個測定項目與異常旗標DTF之值存在相關關係,重要度Q1~QN表示N個測定項目各者對異常旗標DTF之值造成之影響之大小。例如,作為重要度Q1~QN,可採用「對異常旗標DTF之值之貢獻度」。然而,重要度Q1~QN並非限於「貢獻度」者,亦可為例如「貢獻率」等、表示對異常旗標DTF之值造成之影響之大小之其他指標。Returning to FIG. 1, the measurement item
算出重要度Q1~QN之方法可為例如針對示教資料DT,使用決策樹等機械學習演算法,對異常時示教資料DT1及正常時示教資料DT2執行學習,求出故障對應測定資料DFK所含之各測定項目之重要度,作為其計算結果。又,重要度Q1~QN較佳以其等之合計為特定值、例如「1」之方式預先加以正規化。 圖5係顯示重要度資料DQ之一例之圖。圖示之例中,按照重要度由大至小之順序列出測定項目。The method of calculating the importance Q1~QN can be, for example, for the teaching data DT, using a mechanical learning algorithm such as a decision tree to perform learning on the abnormal teaching data DT1 and the normal teaching data DT2 to obtain the fault corresponding measurement data DFK The importance of each measurement item included is the calculation result. In addition, it is preferable that the importance levels Q1 to QN are normalized in advance so that the total of them becomes a specific value, for example, "1". Figure 5 is a diagram showing an example of importance data DQ. In the example shown in the figure, the measurement items are listed in descending order of importance.
返回至圖1,特徵量選定部105自重要度資料DQ中,按照重要度由大至小之順序擷取M個(其中M<N)測定項目,且選擇擷取出之測定項目相關之測定資料作為用於異常診斷之特徵量。特徵量數M可根據與對應之故障模式DFM(參照圖3)關聯之零件知識、或使用者之指定等設定,但大多為例如「2」~「10」之範圍內。然而,於要求更高之診斷精度之情形時,可進而增加特徵量數M。將為了進行異常診斷而選擇之複數個特徵量稱為選擇特徵量DQS(特徵量,參照圖5)。圖5所示之例中,將特徵量數M設為「3」,選擇「風速_Average」、「發電機軸承後方加速度_Max」、及「機艙加速度_Average」之3個測定項目相關之測定資料作為選擇特徵量DQS。Returning to FIG. 1, the
圖1中,故障模式選擇部106自預先設定之故障模式列表ML(參照圖1)中,基於使用者之操作,選擇成為診斷對象之故障模式DFM。又,異常度算出部107基於選擇特徵量DQS(參照圖5)、與正常時示教資料DT2(參照圖4),製成正常模型,且將該正常模型與測定資料之實測值之距離定量化,作為異常度而算出。例如,異常度算出部107可使用馬氏-田口(Maharanobis-Taguchi法)之統計演算法,基於下式(1)計算診斷對象裝置10之異常度A。下式(1)中,x為選擇特徵量DQS之M維之特徵量矢量,μ為正常時示教資料DT2中之特徵量矢量x之平均值,σ為正常時示教資料DT2中之特徵量矢量x之離散度。 In FIG. 1, the failure
裝置狀態判定部108基於異常度A之值,就故障模式DFM進行診斷對象裝置10之狀態為正常或異常之判定。例如,關於故障模式DFM,若異常度A為特定閾值A_th以上,則可判定為「異常」,若異常度A未達閾值A_th,則可判定為「正常」。裝置狀態判定部108將就故障模式DFM判定診斷對象裝置10之狀態是否正常之結果作為狀態資料DC(參照圖1)儲存於資料庫110。即,狀態資料DC為表示針對故障模式DFM之各者,診斷對象裝置10是否正常之資料。The device
圖6係顯示異常度A之時間分佈之一例之圖。圖6中,橫軸為時間,縱軸為異常度A。圖示之例中,由於存在異常度A為閾值A_th以上之時序,故裝置狀態判定部108針對對應之故障模式DFM,判定為「診斷對象裝置10異常」。該裝置狀態判定部108之處理可以如下式(2)所示之偽碼表記。
IF(異常度A≧閾值A_th之時序存在?)
THEN 診斷對象裝置10異常。
ELSE 診斷對象裝置10正常。…式(2)Fig. 6 is a diagram showing an example of the time distribution of anomaly degree A. In Fig. 6, the horizontal axis is time, and the vertical axis is abnormality A. In the example shown in the figure, since there is a sequence in which the degree of abnormality A is greater than or equal to the threshold value A_th, the device
<第1實施形態之動作>
接著,對第1實施形態之動作進行說明。
圖7係異常診斷裝置100中執行之異常診斷處理程序之流程圖。此處,異常診斷處理程序分類成在計算異常度A前執行之離線處理程序R10、與計算異常度A之線上處理程序R20。<Operation of the first embodiment>
Next, the operation of the first embodiment will be described.
FIG. 7 is a flowchart of an abnormality diagnosis processing program executed in the
離線處理程序R10中,處理進行至步驟S12時,示教資料製成部103基於測定資料DM與故障資料DF,製成對應於各個故障模式DFM(參照圖3)之示教資料DT。接著,處理進行至步驟S14時,測定項目重要度算出部104對示教資料DT進行決策樹學習,製成重要度資料DQ。接著,處理進行至步驟S16時,特徵量選定部105針對各個故障模式DFM,對選擇特徵量DQS進行選擇,並將選擇結果儲存於資料庫110。In the offline processing program R10, when the processing proceeds to step S12, the teaching
又,線上處理程序R20中,處理進行至步驟S22時,故障模式選擇部106基於使用者之操作,選擇成為診斷對象之故障模式DFM。接著,處理進行至步驟S24時,異常度算出部107基於選擇特徵量DQS(參照圖5)、正常時示教資料DT2(參照圖4)及上述之式(1),算出異常度A之時間序列分佈,即各取樣時刻之異常度A。接著,處理進行至步驟S26時,裝置狀態判定部108基於異常度A之時間序列分佈,針對對應之故障模式DFM判定診斷對象裝置10有無異常,並根據該判定結果,更新狀態資料DC(參照圖1)。In addition, in the online processing program R20, when the process proceeds to step S22, the failure
<變化例> 本發明並非限定於上述之實施形態者,亦可進行各種變化。上述之實施形態係為了易於理解本發明地進行說明而例示者,未必限定於具備所說明之所有構成者。又,可對上述實施形態之構成追加其他構成,亦可針對構成之一部分而置換成其他構成。又,圖中所示之控制線或資訊線係顯示認為說明上必要者,未必顯示製品上必要之所有控制線或資訊線。可認為實際上幾乎所有構成皆相互連接。針對上述實施形態可能之變化為例如如以下者。<Examples of changes> The present invention is not limited to the above-mentioned embodiment, and various modifications can be made. The above-mentioned embodiments are exemplified in order to facilitate the understanding of the present invention, and are not necessarily limited to those having all the constitutions described. In addition, other structures may be added to the structure of the above-mentioned embodiment, or a part of the structure may be replaced with another structure. In addition, the control lines or information lines shown in the figure show those deemed necessary for explanation, and may not show all the control lines or information lines that are necessary on the product. It can be considered that almost all the components are actually connected to each other. The possible changes to the above-mentioned embodiment are as follows, for example.
(1)上述實施形態中,已說明應用風力發電裝置作為診斷對象裝置10之例。然而,診斷對象裝置10並非限定於風力發電裝置,亦可應用工業機械、電動汽車、鐵路車輛、船舶、廂式電梯、自動扶梯等各種機器作為診斷對象裝置10。(1) In the above-mentioned embodiment, an example in which a wind turbine generator is applied as the
(2)上述實施形態中,為了算出分別對應於N個測定項目之重要度Q1~QN,採用決策樹學習之演算法。然而,為了算出重要度Q1~QN,亦可應用決策樹以外之機械學習演算法。例如,亦可以隨機森林或支持矢量機等演算法學習示教資料DT,求出重要度Q1~QN。(2) In the above embodiment, in order to calculate the importance Q1 to QN corresponding to the N measurement items, the decision tree learning algorithm is used. However, in order to calculate the importance Q1~QN, it is also possible to apply a machine learning algorithm other than the decision tree. For example, algorithms such as random forest or support vector machine may learn the teaching data DT to obtain the importance levels Q1 to QN.
(3)上述實施形態中,為了進行診斷對象裝置10之異常診斷,使用自診斷對象裝置10自身取得之故障資料DF(參照圖3)。然而,例如亦有於剛設置診斷對象裝置10之後尚無故障資料DF、或故障資料DF之量過少之情形。因此,作為故障資料DF,亦可應用與診斷對象裝置10同一規格或相似規格之其他機器中之故障資料DF,基於該資料而產生示教資料DT等。(3) In the above-mentioned embodiment, in order to perform the abnormality diagnosis of the
尤其,診斷對象裝置10為風力發電裝置之情形時,故障資料DF之沿用來源即「其他機器」較佳為相同風力電廠之其他號機,這是因為診斷對象裝置10與「其他機器」之風速、風向、氣溫等自然條件近似之故。又,亦可基於故障資料DF之沿用來源之「其他機器」、與診斷對象裝置10之特性差異,而修正故障資料DF或示教資料DT等。In particular, when the
(4)又,上述實施形態中,由1台異常診斷裝置100實現1個資料庫110。然而,亦可將複數台異常診斷裝置100連接於網路(未圖示),於該網路上之儲存裝置中實現資料庫110。又,亦可藉由以網路上之複數台電腦進行分散處理而實現異常診斷裝置100之功能。(4) In addition, in the above-mentioned embodiment, one
(5)由於上述實施形態中之異常診斷裝置100之硬體可由一般電腦實現,故可將圖7所示之流程圖、執行其他上述之各種處理之程式等儲存於記憶媒體,或經由傳送路徑分發。(5) Since the hardware of the
(6)圖7所示之處理、其他上述之各處理,已於上述實施形態中作為使用程式之軟體式處理進行說明,但亦可將其一部分或全部置換成使用ASIC(Application Specific Integrated Circuit:特殊應用積體電路)、或FPGA(Field Programmable Gate Array:場可程式化閘陣列)等之硬體式處理。(6) The processing shown in Fig. 7 and the other processings described above have been described as software processing using a program in the above embodiment, but part or all of them can be replaced with ASIC (Application Specific Integrated Circuit): Special application integrated circuit), or FPGA (Field Programmable Gate Array: Field Programmable Gate Array) and other hardware processing.
<第1實施形態之效果>
如以上所示,本實施形態之異常診斷裝置100具備:故障模式選擇部106,其選擇成為檢測對象之故障模式DFM;示教資料製成部103,其基於診斷對象裝置10或其他裝置中之與故障模式DFM對應之資料即故障對應測定資料DFK,製成用以判定診斷對象裝置10有無故障之示教資料DT;測定項目重要度算出部104,其基於示教資料DT,算出故障模式DFM中之測定項目P1~PN之重要度Q1~QN;特徵量選定部105,其基於算出之重要度Q1~QN,選擇測定項目P1~PN之一部分作為對於故障模式DFM之特徵量(DQS);異常度算出部107,其基於特徵量(DQS)相關之測定資料DM,算出對應於故障模式DFM之異常度A;及裝置狀態判定部108,其基於算出之異常度A,判定診斷對象裝置10之狀態。<Effects of the first embodiment>
As described above, the
根據本實施形態,由於基於算出之重要度Q1~QN,選擇測定項目P1~PN之一部分作為對於故障模式DFM之特徵量(DQS),故可正確地判定診斷對象裝置10之狀態。According to the present embodiment, since a part of the measurement items P1 to PN is selected as the characteristic quantity (DQS) for the failure mode DFM based on the calculated importance levels Q1 to QN, the state of the
又,故障對應測定資料DFK較佳為自診斷對象裝置10取得之資料。
這是因為認為自診斷對象裝置10取得之故障對應測定資料DFK符合診斷對象裝置10之狀態之程度較高之故。In addition, it is preferable that the failure-corresponding measurement data DFK is data acquired from the
又,示教資料DT包含:推定為發生異常之異常時示教資料DT1、與推定為正常之正常時示教資料DT2;示教資料製成部103選擇自故障發現時日DFT起至往前第1特定時間(T1)之故障對應測定資料DFK,作為異常時示教資料DT1,且選擇自異常時示教資料DT1中最早(ts_c)之資料之上一個(ts_b)資料起至往前第2特定時間(T2)的故障對應測定資料DFK,作為正常時示教資料DT2,第1及第2特定時間(T1、T2)較佳為根據故障模式DFM而設定之時間。In addition, the teaching data DT includes: the teaching data DT1 when the abnormality is presumed to be abnormal, and the teaching data DT2 when the normal is presumed to be normal; the teaching
如此,藉由根據故障模式DFM設定第1及第2特定時間(T1、T2),可設定與診斷對象裝置10之故障狀態相應之適當之第1及第2特定時間(T1、T2)。In this way, by setting the first and second specific times (T1, T2) according to the failure mode DFM, it is possible to set appropriate first and second specific times (T1, T2) corresponding to the failure state of the
1:異常診斷系統 10:診斷對象裝置 20:感測器部 22-1~22-N:感測器 100:異常診斷裝置(電腦) 101:測定資料取得部 102:故障資料取得部 103:示教資料製成部(示教資料製成機構) 104:測定項目重要度算出部(測定項目重要度算出機構) 105:特徵量選定部(特徵量選定機構) 106:故障模式選擇部(故障模式選擇機構) 107:異常度算出部(異常度算出機構) 108:裝置狀態判定部(裝置狀態判定機構) 110:資料庫(資料庫機構) A:異常度 A_th:閾值 DC:狀態資料 DF:故障資料 DFK:故障對應測定資料 DFM:故障模式 DFT:故障發現時日 DM:測定資料 DQ:重要度資料 DQS:選擇特徵量(特徵量) DT:示教資料 DT1:異常時示教資料 DT2:正常時示教資料 ML:故障模式列表 P1~PN:測定項目 Q1~QN:重要度 R10:離線處理程序 R20:線上處理程序 S12:步驟 S14:步驟 S16:步驟 S22:步驟 S24:步驟 S26:步驟 T1:特定時間(第1特定時間) T2:特定時間(第2特定時間) ts_1~ts_max:取樣時刻 ts_a~ts_d:取樣時刻1: Abnormal diagnosis system 10: Diagnostic target device 20: Sensor section 22-1~22-N: Sensor 100: Abnormal diagnosis device (computer) 101: Measurement data acquisition department 102: Failure data acquisition department 103: Teaching material preparation department (teaching material preparation organization) 104: Measurement item importance calculation unit (measurement item importance calculation mechanism) 105: Feature selection part (feature selection mechanism) 106: Failure mode selection section (failure mode selection mechanism) 107: Abnormal degree calculation unit (abnormal degree calculation mechanism) 108: Device status judging section (device status judging mechanism) 110: database (database institution) A: Abnormality A_th: threshold DC: Status data DF: fault data DFK: Failure corresponding measurement data DFM: failure mode DFT: Time when the fault was discovered DM: Measurement data DQ: Importance data DQS: select feature quantity (feature quantity) DT: Teaching data DT1: Teaching data when abnormal DT2: Normal teaching data ML: List of failure modes P1~PN: Measurement items Q1~QN: Importance R10: Offline processing program R20: Online processing program S12: steps S14: Step S16: steps S22: Step S24: steps S26: Step T1: specific time (1st specific time) T2: specific time (second specific time) ts_1~ts_max: sampling time ts_a~ts_d: sampling time
圖1係較佳之第1實施形態之異常診斷系統之方塊圖。 圖2係顯示測定資料之資料構造之一例之圖。 圖3係顯示故障資料之一例之圖。 圖4係顯示示教資料之一例之圖。 圖5係顯示重要度資料之一例之圖 圖6係顯示異常度之時間分佈之一例之圖。 圖7係異常診斷裝置中執行之異常診斷處理程序之流程圖。Fig. 1 is a block diagram of the abnormality diagnosis system of the preferred first embodiment. Figure 2 is a diagram showing an example of the data structure of the measurement data. Figure 3 is a diagram showing an example of fault data. Figure 4 is a diagram showing an example of teaching data. Figure 5 is a diagram showing an example of importance data Figure 6 is a diagram showing an example of the time distribution of abnormality. Fig. 7 is a flowchart of the abnormal diagnosis processing program executed in the abnormal diagnosis device.
1:異常診斷系統1: Abnormal diagnosis system
10:診斷對象裝置10: Diagnostic target device
20:感測器部20: Sensor section
22-1~22-N:感測器22-1~22-N: Sensor
100:異常診斷裝置(電腦)100: Abnormal diagnosis device (computer)
101:測定資料取得部101: Measurement data acquisition department
102:故障資料取得部102: Failure data acquisition department
103:示教資料製成部(示教資料製成機構)103: Teaching material preparation department (teaching material preparation organization)
104:測定項目重要度算出部(測定項目重要度算出機構)104: Measurement item importance calculation unit (measurement item importance calculation mechanism)
105:特徵量選定部(特徵量選定機構)105: Feature selection part (feature selection mechanism)
106:故障模式選擇部(故障模式選擇機構)106: Failure mode selection section (failure mode selection mechanism)
107:異常度算出部(異常度算出機構)107: Abnormal degree calculation unit (abnormal degree calculation mechanism)
108:裝置狀態判定部(裝置狀態判定機構)108: Device status judging section (device status judging mechanism)
110:資料庫(資料庫機構)110: database (database institution)
A:異常度A: Abnormality
DC:狀態資料DC: Status data
DF:故障資料DF: fault data
DM:測定資料DM: Measurement data
DQ:重要度資料DQ: Importance data
DT:示教資料DT: Teaching data
ML:故障模式列表ML: List of failure modes
Claims (4)
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2019232052A JP7281394B2 (en) | 2019-12-23 | 2019-12-23 | Abnormal diagnosis device and program |
JP2019-232052 | 2019-12-23 |
Publications (2)
Publication Number | Publication Date |
---|---|
TW202125138A true TW202125138A (en) | 2021-07-01 |
TWI762101B TWI762101B (en) | 2022-04-21 |
Family
ID=76541283
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW109145508A TWI762101B (en) | 2019-12-23 | 2020-12-22 | Abnormal Diagnosis Device and Abnormal Diagnosis Program |
Country Status (2)
Country | Link |
---|---|
JP (1) | JP7281394B2 (en) |
TW (1) | TWI762101B (en) |
Family Cites Families (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7389204B2 (en) * | 2001-03-01 | 2008-06-17 | Fisher-Rosemount Systems, Inc. | Data presentation system for abnormal situation prevention in a process plant |
US7079984B2 (en) * | 2004-03-03 | 2006-07-18 | Fisher-Rosemount Systems, Inc. | Abnormal situation prevention in a process plant |
US8762106B2 (en) * | 2006-09-28 | 2014-06-24 | Fisher-Rosemount Systems, Inc. | Abnormal situation prevention in a heat exchanger |
CN102478825A (en) * | 2010-11-23 | 2012-05-30 | 大连兆阳软件科技有限公司 | Implementation method of remote monitoring and fault diagnosis system of numerical control machine |
WO2015005663A1 (en) * | 2013-07-10 | 2015-01-15 | 주식회사 글로비즈 | Signal measurement diagnosis monitoring system and method therefor, and method and system for applying same to individual device |
CN107850464A (en) * | 2016-07-13 | 2018-03-27 | 株式会社五十岚电机制作所 | Rotation angle detection apparatus and the electro-motor for being provided with the rotation angle detection apparatus |
CN106444703B (en) * | 2016-09-20 | 2018-12-07 | 西南石油大学 | Dynamic equipment running status fuzzy evaluation and prediction technique based on fault mode probability of happening |
US11669771B2 (en) * | 2017-07-13 | 2023-06-06 | Nec Corporation | Learning system, analysis system, learning method, and storage medium |
WO2019049688A1 (en) * | 2017-09-06 | 2019-03-14 | 日本電信電話株式会社 | Abnormal sound detecting device, abnormality model learning device, abnormality detecting device, abnormal sound detecting method, abnormal sound generating device, abnormal data generating device, abnormal sound generating method, and program |
WO2019107315A1 (en) * | 2017-11-28 | 2019-06-06 | 国立研究開発法人産業技術総合研究所 | Method and system for detecting symptom of abnormality in apparatus being monitored |
JP7006282B2 (en) * | 2018-01-12 | 2022-01-24 | 株式会社明電舎 | Equipment abnormality diagnostic equipment |
CN108490284B (en) * | 2018-02-12 | 2021-10-15 | 国网山东省电力公司电力科学研究院 | New energy data acquisition device, system and method for multiple application scenes |
JP7320368B2 (en) * | 2019-04-09 | 2023-08-03 | ナブテスコ株式会社 | FAILURE PREDICTION DEVICE, FAILURE PREDICTION METHOD AND COMPUTER PROGRAM |
CN110531656A (en) * | 2019-08-13 | 2019-12-03 | 大唐水电科学技术研究院有限公司 | A kind of monitoring system and method for Hydropower Unit performance |
-
2019
- 2019-12-23 JP JP2019232052A patent/JP7281394B2/en active Active
-
2020
- 2020-12-22 TW TW109145508A patent/TWI762101B/en active
Also Published As
Publication number | Publication date |
---|---|
TWI762101B (en) | 2022-04-21 |
JP2021099744A (en) | 2021-07-01 |
JP7281394B2 (en) | 2023-05-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP3575892B1 (en) | Model parameter value estimation device and estimation method, program, recording medium with program recorded thereto, and model parameter value estimation system | |
KR101903283B1 (en) | Automatic diagnosis system and automatic diagnosis method | |
WO2017154844A1 (en) | Analysis device, analysis method, and analysis program | |
JP3993825B2 (en) | Inference signal generator for instrumented equipment and processes | |
KR101955305B1 (en) | Gas turbine sensor failure detection utilizing a sparse coding methodology | |
US9122273B2 (en) | Failure cause diagnosis system and method | |
CN102265227B (en) | Method and apparatus for creating state estimation models in machine condition monitoring | |
US20130060524A1 (en) | Machine Anomaly Detection and Diagnosis Incorporating Operational Data | |
CN107609574A (en) | Wind turbines fault early warning method based on data mining | |
EP3444724B1 (en) | Method and system for health monitoring and fault signature identification | |
WO2018171165A1 (en) | Fault prediction method and device for fan | |
US12007753B2 (en) | System and method for predicting industrial equipment motor behavior | |
CN108956111B (en) | Abnormal state detection method and detection system for mechanical part | |
Mosallam et al. | Component based data-driven prognostics for complex systems: Methodology and applications | |
WO2016195092A1 (en) | Anomaly sensing device | |
CN112884199A (en) | Method and device for predicting faults of hydropower station equipment, computer equipment and storage medium | |
JP2018163645A (en) | Fault diagnosis device, monitoring device, fault diagnosis method, and fault diagnosis program | |
CN116086537A (en) | Equipment state monitoring method, device, equipment and storage medium | |
TWI780434B (en) | Abnormal diagnosis device and method | |
KR102108975B1 (en) | Apparatus and method for condition based maintenance support of naval ship equipment | |
JP5949032B2 (en) | Pre-processing method and abnormality diagnosis device | |
TWI762101B (en) | Abnormal Diagnosis Device and Abnormal Diagnosis Program | |
CN108629077B (en) | Fault diagnosis device, monitoring device, fault diagnosis method, and recording medium | |
US20180087489A1 (en) | Method for windmill farm monitoring | |
US11495114B2 (en) | Alert similarity and label transfer |