TWI564554B - A real-time xxy platform fault diagnosis method - Google Patents
A real-time xxy platform fault diagnosis method Download PDFInfo
- Publication number
- TWI564554B TWI564554B TW103143554A TW103143554A TWI564554B TW I564554 B TWI564554 B TW I564554B TW 103143554 A TW103143554 A TW 103143554A TW 103143554 A TW103143554 A TW 103143554A TW I564554 B TWI564554 B TW I564554B
- Authority
- TW
- Taiwan
- Prior art keywords
- platform
- motor
- fault
- xxy
- signal
- Prior art date
Links
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Description
本發明是有關於一種檢測方法，特別是一種用於XXY平台即時故障之檢測方法。 The invention relates to a detection method, in particular to a detection method for an immediate fault of an XXY platform.
由於各產業的技術發達，使現今產品多走向發展精密化，使其產品空間需求減小，甚至是隨身攜帶，但相對使產品內部組件更加縮小，組件安裝上難度增加，為了將上述所提各微小組件進行組裝，在組裝上需要更高的精確度，因此以人工方式進行組裝已無法承擔，並造成生產速度降低，因此對位平台的出現，在有限的條件下改善了人工定位的方式。 Due to the developed technology of various industries, the current products are more developed and refined, so that their product space needs are reduced, even if they are carried around, but the internal components of the products are further reduced, and the difficulty in component installation is increased. The assembly of small components requires higher precision in assembly, so assembly by manual means cannot be undertaken, and the production speed is reduced. Therefore, the appearance of the alignment platform improves the manual positioning under limited conditions.
現在在對位的需求上，對位平台不僅需要其單方向的準確度，在角度上也需要作修正，才可符合現在之對位需求，因此開發XYθ平台，提供較佳的對位準確度，但有高度較高的缺點，因此空間需求較高，並且因為由馬達控制平台中心點轉動角度的關係，在耐重上也有所限制，因此出現一種與XYθ平台一樣藉由3顆馬達進行定位的XXY平台，其在單軸移動上與XYθ平台類似，不同的是X軸移動是藉由X1、X2馬達進行同步的方式進行，而在角度轉動上則藉由 X1、X2馬達前後交錯的方式，使其在角度上發生變動，最後再藉由Y馬達移動，將平台中心點誤差進行矯正，藉此方式進行角度轉動。 Nowadays, in terms of alignment, the alignment platform not only needs its unidirectional accuracy, but also needs to be corrected in terms of angle to meet the current alignment requirements. Therefore, the XYθ platform is developed to provide better alignment accuracy. However, there is a high degree of disadvantage, so the space requirement is high, and because of the rotation angle of the center point of the motor control platform, there is also a limitation in the weight resistance, so that there is a positioning by three motors like the XYθ platform. The XXY platform is similar to the XYθ platform in uniaxial movement, except that the X-axis movement is performed by the X1 and X2 motors, while the angular rotation is performed by the angle. The X1 and X2 motors are staggered in front and rear to change the angle. Finally, the Y-motor is moved to correct the error of the center point of the platform, and the angle is rotated.
XXY平台與XYθ平台兩者相較下，XXY平台改善XYθ平台體積過大與耐重量不足的缺點，但當XXY平台發生故障時，對其XXY平台故障檢測的研究資料較少，相對於XYθ平台更難以檢測，在XXY平台上常見故障分成三大類，分別為馬達故障，驅動器故障與平台故障，而本發明中以XXY平台馬達故障作為主題，對此進行故障檢測；在對各類故障檢測的論文中，對待測平台故障進行特徵訊號提取，其主要分為以各種機械故障作為檢測訊號、或以電流、電壓訊號，多作為電氣的故障檢測檢測訊號、或以混合複數訊號源互相搭配作為故障檢測，而在檢測方法上多數配合上述所提之訊號進行訊號分析，在訊號分析上可用多種方式進行，如以小波分析進行故障檢測、頻譜分析進行檢測或類神經進行檢測等等；然，就以小波分析舉例來說，其於小波轉換中萃取的特徵數明顯較多，加上XXY平台上，往往存在許多雜訊的干擾，使得過去所發展的演算方法在受雜訊干擾時作明確之辨識，導致監測準確率降低，實須改善。 Compared with the XXY platform and the XYθ platform, the XXY platform improves the XYθ platform's excessive volume and insufficient weight resistance. However, when the XXY platform fails, there is less research data on the XXY platform fault detection, which is more than the XYθ platform. It is difficult to detect. Common faults on the XXY platform are divided into three categories, namely motor fault, drive fault and platform fault. In the present invention, the fault of the XXY platform motor is the subject of fault detection; in the papers on various fault detection In the process of extracting characteristic signals from the faults to be tested, it is mainly divided into various mechanical faults as detection signals, or current and voltage signals, as electrical fault detection and detection signals, or mixed with multiple signal sources as fault detection. Most of the detection methods are combined with the above-mentioned signals for signal analysis, and the signal analysis can be performed in various ways, such as fault detection by wavelet analysis, spectrum analysis for detection, or nerve-like detection, etc.; For example, wavelet analysis has significantly more features extracted in wavelet transform. XXY on the platform, there is often a lot of noise interference, making the calculation method in the past for the development of clear identification at by noise interference, resulting in reduced monitoring accuracy, real need to improve.
因此，本發明之目的，是在提供一種用於XXY平台即時故障之檢測方法，其具有方法簡單、快速監測的特 點，使其檢測方法可以提供較佳的即時故障診斷環境之效果。 Therefore, the object of the present invention is to provide a method for detecting an immediate fault of an XXY platform, which has a simple method and a fast monitoring method. Point, so that its detection method can provide a better immediate fault diagnosis environment.
因此，本發明用於XXY平台即時故障之檢測方法，其包含有讀取振動訊號、濾波步驟、混沌訊號同步、可拓理論分析、可拓集合運算等步驟，是以，透過XXY平台所產生的微小振動量讀取各馬達振動訊號，進行濾波步驟已分別得知各馬達之提取訊號後，將提取訊號透過混沌訊號同步檢測方式，即將提取訊號分別輸入僕系統與主系統內，使XXY平台各狀態間的差異性放大，並將主、僕混沌兩系統相減後而得知其動態誤差，而後再透過該可拓理論分析、可拓集合運算等步驟，以計算各狀態之動態誤差作為特徵值，並透過可拓運算判別狀態間的關聯函數，藉以判斷XXY平台為何種故障狀態，以有效縮短演算與判斷時間，使該檢測方法可以提供較佳的即時故障診斷環境外，並且更具有簡單、快速的進行監測之功效。 Therefore, the present invention is applied to a method for detecting an immediate fault of an XXY platform, which comprises the steps of reading a vibration signal, filtering step, chaotic signal synchronization, extension theory analysis, extension set operation, etc., which are generated by the XXY platform. The vibration amount of each motor is read by the small vibration amount, and the filtering step has separately obtained the extracted signals of the motors, and then the extracted signals are transmitted through the chaotic signal synchronous detection mode, that is, the extracted signals are respectively input into the servant system and the main system, so that the XXY platform is respectively The difference between states is magnified, and the main and servant chaotic systems are subtracted to know the dynamic error, and then the extension theory analysis and the extension set operation are used to calculate the dynamic error of each state as a feature. Value, and through the extension operation to determine the correlation function between states, to determine the fault state of the XXY platform, in order to effectively shorten the calculation and judgment time, so that the detection method can provide a better immediate fault diagnosis environment, and is more simple Quickly monitor the effect.
圖1是本發明一較佳實施例之方塊圖。 1 is a block diagram of a preferred embodiment of the present invention.
圖2平台正常振動訊號。 Figure 2 shows the normal vibration signal of the platform.
圖3平台正常振動訊號經FFT。 Figure 3 shows the normal vibration signal of the platform via FFT.
圖4平台正常振動訊號經FFT放大圖。 Figure 4 shows the normal vibration signal of the platform through the FFT magnified view.
圖5平台正常振動訊號經FFT放大圖。 Figure 5 shows the normal vibration signal of the platform through the FFT magnified view.
圖6提取X2馬達訊號圖。 Figure 6 extracts the X2 motor signal map.
圖7提取Y馬達訊號圖。 Figure 7 extracts the Y motor signal map.
圖8提取X1馬達訊號經混沌訊號同步軌跡圖。 Figure 8 extracts the X1 motor signal through the chaotic signal synchronization trace map.
圖9提取X1馬達訊號經混沌訊號同步軌跡圖。 Figure 9 extracts the X1 motor signal through the chaotic signal synchronization trace map.
圖10提取Y馬達訊號經混沌訊號同步軌跡圖。 Figure 10 extracts the trajectory of the Y motor signal through the chaotic signal synchronization track.
圖11 X1馬達平台故障振動訊號。 Figure 11 X1 motor platform fault vibration signal.
圖12 X1馬達平台故障振動訊號經FFT。 Figure 12 X1 motor platform fault vibration signal is FFT.
圖13 X1馬達故障之提取X1訊號。 Figure 13 X1 signal extraction X1 signal.
圖14 X1馬達故障之X1提取訊號動態誤差圖。 Figure 14 X1 extraction signal dynamic error diagram of X1 motor failure.
圖15 X2馬達故障之X2提取訊號動態誤差圖示平台之狀態。 Figure 15 X2 Motor Fault X2 Extract Signal Dynamic Error Graphical Status of the Platform.
圖16 Y馬達故障之Y提取訊號動態誤差圖。 Figure 16 Y motor fault Y extraction signal dynamic error map.
附件一 使用dSPACE介面所建立的XXY平台監控診斷介面。 Annex I The monitoring interface is monitored by the XXY platform established by the dSPACE interface.
有關本發明之前述及其他技術內容、特點與功效，在以下配合參考圖式之較佳實施例的詳細說明中，將可清楚的明白。 The above and other technical contents, features, and advantages of the present invention will become apparent from the Detailed Description of the <RTIgt;
參閱圖1，本發明一較佳實施例，其步驟包含有讀取振動訊號、濾波步驟、混沌訊號同步、可拓理論分析、可拓集合運算；其中，該讀取振動訊號即偵測XXY平台(即待測平台)中的X1馬達、X2馬達、Y馬達所產生之振動訊號，而本發明中感測器數量與裝設，希望使用較少的感測器進行故障檢測，最後在訊號選擇上以振動訊號方式作為檢測平台 故障訊號，並且透過一加速規(Accelerometer)設置於移動平台上，以進行平台振動量的感測而得到X1馬達、X2馬達及Y馬達之振動訊號；另，該濾波步驟中，其將前述振動訊號進行濾波步驟，以分別得知X1馬達、X2馬達及Y馬達之提取訊號M (t)及S (t)。 Referring to FIG. 1, in a preferred embodiment of the present invention, the steps include: reading a vibration signal, a filtering step, a chaotic signal synchronization, an extension theory analysis, and an extension set operation; wherein the reading the vibration signal detects the XXY platform. (ie, the vibration signal generated by the X1 motor, the X2 motor, and the Y motor in the platform to be tested), and the number and arrangement of the sensors in the present invention, it is desirable to use fewer sensors for fault detection, and finally in the signal selection The vibration signal is used as the detection platform failure signal, and is set on the mobile platform through an Accelerometer to sense the vibration amount of the platform to obtain the vibration signals of the X1 motor, the X2 motor and the Y motor; In the filtering step, the filtering signal is filtered to obtain the extracted signals M ( t ) and S ( t ) of the X1 motor, the X2 motor and the Y motor, respectively.
仍續前述，該混沌訊號同步步驟中，其主要具有一主系統與一僕系統，將該等提取訊號M (t)及S (t)分別輸入主、僕系統內，而該主系統(1)、僕系統(2)定義如下：
此外，將主、僕混沌兩系統作相減(如式3)，即可得到動態誤差狀態：
其中，E 1=Xm 1-Xs 1，E 2=Xm 2-Xs 2，E n =Xm n -Xs n Where E 1 = Xm 1 - Xs 1 , E 2 = Xm 2 - Xs 2 , E n = Xm n - Xs n
最後動態誤差狀態也會產生混沌現象，因此利用混沌現象之運動軌跡，增加各平台狀態間產生之振動訊號差
別，藉由混沌動態誤差方程式作為辨識的依據，而本實施中該混沌系統可使用羅倫茲混沌系統(Lorenz Chaos System)，將兩個相同的羅倫茲系統分別區分為主系統(4)與僕系統(5)，分別如下：
本實施例將提取訊號M (t)輸入作提取並送入主系統中，主系統輸入訊號方式如下：x m1=M(t),x m2=M (t+1),x m3=M (t+2) In this embodiment, the extracted signal M ( t ) input is extracted and sent to the main system, and the main system inputs the signal as follows: x m 1 =M ( t ) , x m 2 = M ( t +1) , x m 3 = M ( t +2)
本實施例將提取訊號S (t)輸入作提取並送入僕系統中，僕系統輸入訊號方式如式(6)：x s1=S (t),x s2=S (t+1),x s3=S (t+2) (6) In this embodiment, the extracted signal S ( t ) is input and sent to the servant system, and the servant system inputs the signal as in equation (6): x s 1 = S ( t ) , x s 2 = S ( t +1) , x s 3 = S ( t +2) (6)
最後動態誤差系統以矩陣型式則可表示如式(7)
其中，E 1=x m1-x s1，E 2=x m2-x s2，E 3=x m3-x s3最後得到的動態誤差E1、E2與E3，繪出動態誤差軌跡圖作為觀察，藉此獲得各動態誤差值之重心值，藉此減少運算量並增 加辨別度，最後透過可拓理論檢測待測平台對各狀態之關聯度。 Where E 1 = x m 1 - x s 1 , E 2 = x m 2 - x s 2 , E 3 = x m 3 - x s 3 The resulting dynamic errors E1, E2 and E3, plot the dynamic error trajectory As an observation, the graph obtains the center-of-gravity value of each dynamic error value, thereby reducing the amount of calculation and increasing the degree of discrimination. Finally, the degree of association between the states of the platform to be tested is detected by the extension theory.
至於，該可拓理論分析步驟中，則是透過動態誤差E1、E2、E3縮小的變化作為故障特徵，並且以動態誤差E1、E2、E3之其一故障狀態建立一物元模型，並將該物元模型的動態誤差E1、E2、E3的變動範圍作為一經典域範圍<x1、x2>，將建立物元特徵中各狀態經典域之最大值與最小值作為一節域範圍<y1、y2>，而本實施例中，該物元模型可包括有正常狀態、X1馬達故障、X2馬達故障、Y馬達故障等；而本實施例中，物元理論可以表示物元間的不同，可藉由顯示物件的名稱、特徵名稱與其特徵，將事物做區分，表示各物件間差異，物元模型定義如下：R=(N,C,V) (8) As for the extension theory analysis step, the change that is reduced by the dynamic errors E1, E2, E3 is taken as a fault feature, and a matter-element model is established by one of the dynamic errors E1, E2, E3, and the matter model is The variation range of the dynamic error E1, E2, and E3 of the matter-element model is a classic domain range <x1, x2>, and the maximum and minimum values of the classical domains of each state in the matter-element feature are established as a domain range <y1, y2>. In this embodiment, the matter element model may include a normal state, an X1 motor fault, an X2 motor fault, a Y motor fault, etc.; in this embodiment, the matter-element theory may indicate a difference between the matter elements, Display the name, feature name and characteristics of the object, distinguish the things, and indicate the difference between the objects. The matter element model is defined as follows: R = ( N , C , V ) (8)
而前述中N為物元的名稱，C為物元特徵，V為物元特徵值，在單一的物件中，可以擁有複數個物元特徵，且特徵值若為一區域<x1、x2>，此區域稱為經典域，即經典域為物元特徵值的範圍，所有經典域需在節域<y1、y2>之中，即節域包含該特徵中各狀態經典域的範圍。 Wherein N is the name of the matter element, C is the matter element feature, and V is the matter element feature value. In a single object, there may be a plurality of matter element features, and if the feature value is a region <x1, x2>, This area is called the classical domain, that is, the classical domain is the range of the feature values of the matter elements. All the classical domains need to be in the section <y1, y2>, that is, the section contains the range of the classical domains of each state in the feature.
最後，該可拓集合運算步驟中，其是為了可了解待測平台與各物元模型之關聯函數，最後再藉由正規化結果進行判斷，即所謂的可拓集合運算，係將物元特徵值範圍作為集合範圍，並使用待測平台各物元間的關聯函數，當關聯
函數被運算即可得到待測平台對物元的量化，得到其對各物元的可拓關係，藉由可拓關係了解待測平台與何種物元模型較為接近，而當有一待測平台求其對物元經典域的關聯度，則計算方式運算如下：
將式(9)(10)為待測物元定量，v為待測平台之值，透過計算關聯度，即可得到待測平台與經典域以及節域的關係，即可用式(11)運算在待測平台與該經典域之關聯函數如下：
當關聯函數k(v)計算出來後，若為正值，待測平台在經典域中，反之若為負值，待測平台在經典域外，當有多個物元模型下，可依需求進行權重數調整，但注意各狀態權重值相加需為1，並與關聯函數相乘以作權重處理。 When the correlation function k(v) is calculated, if it is positive, the platform to be tested is in the classical domain, and if it is negative, the platform to be tested is outside the classical domain. When there are multiple matter-element models, it can be performed according to requirements. The weight number is adjusted, but note that the sum of the state weight values needs to be 1, and multiplied by the correlation function for weight processing.
最後將透過權重處理後的各物元之關聯後，將其集合<kk(v)min,kk(v)max>藉由正規化公式為<-1,1>以方便進行判斷，正規化算式如式14，當正規化後狀態關聯函數為1者，則表示此待測平台與當前物元狀態最為接近。 Finally, after the correlation of each matter element processed by the weight, the set <kk(v)min, kk(v)max> is judged by the normalization formula as <-1, 1>, and the normalized formula is obtained. As shown in Equation 14, when the state correlation function is one after normalization, it indicates that the platform to be tested is closest to the current matter element state.
換言之，最後將權重處理後的關聯函數<kk(v)min,kk(v)max>，藉由正規化為<-1,1>以方便進行判斷，當正規化後狀態關聯函數為1者，則表示此待測平台目前狀態。 In other words, the weighted processed correlation function <kk(v)min, kk(v)max> is finally determined by normalization to <-1, 1>, and the state correlation function is 1 when normalized. , indicating the current status of the platform under test.
以下將透過實驗方式觀察各狀態資料，建立出各故障狀態之物元模型，並藉由運算該XXY平台對各故障狀態的關聯函數，即可了解該XXY平台之故障狀態。 In the following, each state data is observed experimentally, and the matter-element model of each fault state is established, and the fault state of the XXY platform can be understood by calculating the correlation function of the fault state of the XXY platform.
本實施例進行實驗時，其用於XXY對位平台之故障檢測，以透過加速規作為感測器，對平台振動量進行感測以得到一振動訊號，訊號可使用類比數位轉換器(Analog-to-digital converter，以下簡稱ADC)方式透過dSPACE送至電腦，於電腦端作訊號處理與分析，並透過電腦端顯示其平台波形與故障狀態；每當XXY平台出現不同故障狀態時，會對應產生不同振動量與頻率，因此將此振動訊號提取，並將訊號送入混沌訊號同步系統中的主、僕系統，透過混沌訊號同步使其平台各狀態間的差異性增加，藉此提升平台狀態的辨識度，並透過可拓方法，製作各狀態的物元模型，計算待測平台與物元模型的關係，進而辨別出待測平台之故障狀態；本實驗透過將馬達作斷路動作，將停止的馬達視為故障馬達，藉此模擬平台發生馬達故障，所提可辨識平台狀態分別為正常狀態、X1馬達故障、X2馬達故障與Y馬達故障 共四種狀態。 When the experiment is carried out in this embodiment, it is used for fault detection of the XXY alignment platform to sense the vibration amount of the platform through the acceleration gauge as a sensor to obtain a vibration signal, and the signal can use an analog digital converter (Analog- To-digital converter (hereinafter referred to as ADC) is sent to the computer through dSPACE, and performs signal processing and analysis on the computer side, and displays its platform waveform and fault status through the computer; whenever the XXY platform has different fault conditions, it will be generated accordingly. Different vibration quantities and frequencies, so the vibration signal is extracted, and the signal is sent to the main and servant system in the chaotic signal synchronization system, and the difference between the states of the platform is increased by chaotic signal synchronization, thereby improving the state of the platform. Identification, and through the extension method, the matter element model of each state is produced, the relationship between the platform to be tested and the matter-element model is calculated, and then the fault state of the platform to be tested is identified; the experiment is stopped by the motor breaking action. The motor is regarded as a faulty motor, so that the motor failure occurs on the simulated platform, and the identifiable platform state is normal state, X1 motor Barrier, X2 motor failure and Y motor failure There are four states.
仍續前述，正常狀態下平台之振動訊號如圖2所示，此振動訊號受到來自三顆步進馬達造成的振動所影響，因此正常訊號分別由Y、X1與X2共三顆馬達的振動訊號組成，如此可以先將正常訊號藉由FFT方式進行頻譜分析如圖3所示，藉由FFT了解主要頻段與雜訊的分佈，而三顆馬達的主要訊號頻段位置於500Hz至1000Hz頻段處，再將此處頻段放大如圖4所示，可以發現三個突出振幅於此頻段範圍，經實驗結果發現，突出振幅由左至右分別表示X1馬達、X2馬達、Y馬達所造成的振動量主要頻段。 As mentioned above, the vibration signal of the platform under normal conditions is shown in Fig. 2. The vibration signal is affected by the vibration caused by the three stepping motors. Therefore, the normal signals are respectively composed of three motors of Y, X1 and X2. Composition, so that the normal signal can be first analyzed by FFT method as shown in Fig. 3. The FFT is used to understand the distribution of the main frequency band and the noise, and the main signal frequency bands of the three motors are located in the frequency band of 500 Hz to 1000 Hz. The frequency band is amplified as shown in Fig. 4. Three protruding amplitudes can be found in this frequency range. The experimental results show that the protruding amplitudes from left to right indicate the main frequency of vibration caused by X1 motor, X2 motor and Y motor. .
本發明中提取進入電腦的平台振動訊號，透過simulink建立濾波器(filter)進行濾波(Filtering)，濾波方式選擇巴特沃斯濾波器(Butterworth filter)，主因在於與其他類型濾波器類型相比，衰減速度相對緩慢且十分平坦，透過帶通濾波(band pass)方式突顯各馬達狀態的訊號並作提取，即可得到各馬達的提取訊號如圖5至圖7所示，而此訊號相當微小，因此可將提取出的X1馬達、X2馬達與Y馬達的提取訊號透過羅倫茲混沌訊號同步系統進行特徵放大，分別輸入主、僕系統中，並將該主、僕混沌兩系統相減以得動態誤差E1、E2、E3，而各馬達動態誤差E1、E2、E3的動態誤差圖呈如圖8至圖10所示，動態誤差圖會因混沌吸引子的關係，使圖形會出現繞圈的現象發生。 In the present invention, the platform vibration signal that is input into the computer is extracted, and a filter is established through a simulink filter (Filtering), and a Butterworth filter is selected in a filtering manner. The main reason is that the attenuation is compared with other types of filter types. The speed is relatively slow and very flat. The signal of each motor state is highlighted and extracted by band pass method, and the extracted signals of each motor are obtained as shown in FIG. 5 to FIG. 7, and the signal is relatively small, so The extracted signals of the extracted X1 motor, X2 motor and Y motor can be amplified by the Lorentz chaotic signal synchronization system, input into the main and servant systems, and the main and servant chaotic systems are subtracted to obtain dynamic Errors E1, E2, E3, and the dynamic error diagrams of each motor dynamic error E1, E2, E3 are shown in Figure 8 to Figure 10. The dynamic error map will cause the pattern to appear around the circle due to the chaotic attractor relationship. occur.
仍續前續，以X1馬達故障為例，振動訊號產生改變如圖11，可以發現所提取訊號減小且頻率改變，將訊號經FFT轉換後可以發現變化如圖12，於頻譜600Hz至650Hz處振幅消失，也就表示X1馬達出現故障，此變化透過X1馬達提取量也可以發現如圖13，並將X1馬達提取量透過進入混沌訊號同步系統中，可發現馬達發生故障後，與原先正常狀態下提取訊號所產生動態誤差圖狀態差距更大。 Continued on the continuation, taking the X1 motor fault as an example, the vibration signal changes as shown in Figure 11. It can be found that the extracted signal is reduced and the frequency is changed. After the signal is converted by FFT, the change can be found as shown in Fig. 12, at the spectrum of 600 Hz to 650 Hz. If the amplitude disappears, it means that the X1 motor is faulty. This change can also be found through the X1 motor extraction. As shown in Figure 13, the X1 motor extraction amount is transmitted into the chaotic signal synchronization system, and the motor can be found to be in the normal state after the fault occurs. The state of the dynamic error map generated by the extracted signal is larger.
再者，當其他馬達發生故障，所對應馬達的提取量經過混沌訊號同步後之混沌誤差圖，即如圖14至圖16所示，經過混沌訊號同步方法後，可以看到故障馬達與正常馬達的動態誤差圖產生變化，比起原先未使用混沌訊號同步大上許多，當馬達發生故障，各動態誤差值E1、E2、E3皆會縮小，使其比原先提取訊號更讓人易於觀察與分辨平台狀態，並且使故障檢測辨識率增加；在其他馬達故障狀態下也會發生一樣的改變，當故障發生，其動態誤差圖皆會發生縮小的現象。 Furthermore, when other motors fail, the extracted amount of the corresponding motor is chaotic error map after chaotic signal synchronization, as shown in Fig. 14 to Fig. 16, after the chaotic signal synchronization method, the fault motor and the normal motor can be seen. The dynamic error map produces a change, which is much larger than the synchronization of the original unused chaotic signal. When the motor fails, the dynamic error values E1, E2, and E3 are reduced, making it easier to observe and distinguish than the original extracted signal. The state of the platform, and the recognition rate of the fault detection is increased; the same change occurs in other motor fault states, and when the fault occurs, the dynamic error map will shrink.
仍續前述，透過E1、E2與E3縮小的變化作為故障特徵，將三個動態誤差分別作重心值為特徵值，但可以發現動態誤差圖以0作為中心轉動，因此各狀態重心值會趨近於0，因此可改將動態誤差值以正值與負值各作重心作為特徵，最後將正值重心與負值重心距離作為判斷特徵值，藉此減少計算量；並發現動態誤差E2對平台各故障狀態最為敏
感，因此以E2作為故障狀態的物元模型，將各狀態E2之變動範圍作為經典域範圍，將特徵中各狀態經典域之最大值與最小值作為節域範圍，並將各故障物元模型與節域建立完成如下表：
仍續前述，並將3個特徵值權重皆設為1/3，最後即可使用可拓方法運算，了解待測平台與各物元模型之關聯函數，最後可藉由正規化結果進行判斷，當關聯函數為1時，將讓使用者很明確地知道所屬類別的程度。 The above is continued, and the weights of the three eigenvalues are all set to 1/3. Finally, the extension method can be used to understand the correlation function between the platform to be tested and each matter-element model, and finally, the result can be judged by the normalization result. When the correlation function is 1, it will let the user know the extent of the category very clearly.
再者，本發明中，更可以dSPACE的ADC提取平台振動訊號，ADC對訊號之提取頻率為10Khz，本實施例以每1000筆資料診斷一次平台，也就是每0.1秒判斷一次平台狀態，藉此達到平台即時故障監測與診斷，而電腦端平台診斷dSPACE介面如圖17所示，透過此介面可以觀看波形圖
與關聯函數的變化，訊號圖中分別可看到平台振動訊號，以及由輸入訊號中所提取出代表各馬達狀態之訊號，並且可透過各馬達所顯示之經正規化後之關聯函數，可經由此數值判斷平台故障狀態；在故障檢測準確率上，係以將各故障狀態資料作10000筆測試檢測結果之準確率為主，使用本發明所提之混沌訊號同步可拓方法對平台各狀態檢測之準確率如下表：
由上表可看到平台各狀態故障檢測重心準確率可達到98%，結果證實本發明提出方法能夠有效的診斷出四種平台故障狀態；是以，本發明只需透過單一感測器即可對平台四種狀態作故障檢測，且最後準確率可達98%，本發明提之檢測方法對平台有良好的檢測效果，將各馬達之振動訊號濾波後所得之提取訊號透過混沌訊號同步檢測方式，使其平台各狀態間的差異性增加，藉此提升平台狀態的辨識度，而後透過主、僕系統相減後所得之各狀態之動態誤差作為特徵值，並透過可拓運算判別狀態間的關聯函數，藉以判斷XXY 平台為何種故障狀態，藉此方式縮短演算與判斷時間，此方法具有簡單、快速監測的特點，使其檢測方法可以提供較佳的即時故障診斷環境並達到其目的。 It can be seen from the above table that the accuracy of the center of gravity detection of the platform can reach 98%. The results prove that the proposed method can effectively diagnose the four platform fault states; therefore, the present invention can only be transmitted through a single sensor. The fault detection is performed on the four states of the platform, and the final accuracy rate is up to 98%. The detection method of the present invention has a good detection effect on the platform, and the extracted signal obtained by filtering the vibration signals of each motor is transmitted through the chaotic signal synchronous detection method. The difference between the states of the platform is increased, thereby enhancing the recognition degree of the platform state, and then the dynamic error of each state obtained by subtracting the main and the servant system is used as the feature value, and the state between the states is discriminated by the extension operation. Correlation function to judge XXY The fault state of the platform is used to shorten the calculation and judgment time. This method has the characteristics of simple and rapid monitoring, so that the detection method can provide a better instant fault diagnosis environment and achieve its purpose.
惟以上所述者，僅為說明本發明之較佳實施例而已，當不能以此限定本發明實施之範圍，即大凡依本發明申請專利範圍及發明說明書內容所作之簡單的等效變化與修飾，皆應仍屬本發明專利涵蓋之範圍內。 The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, that is, the simple equivalent changes and modifications made in accordance with the scope of the present invention and the contents of the description of the invention. All should remain within the scope of the invention patent.
Claims (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW103143554A TWI564554B (en) | 2014-12-12 | 2014-12-12 | A real-time xxy platform fault diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW103143554A TWI564554B (en) | 2014-12-12 | 2014-12-12 | A real-time xxy platform fault diagnosis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TW201621292A TW201621292A (en) | 2016-06-16 |
| TWI564554B true TWI564554B (en) | 2017-01-01 |
Family
ID=56755402
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW103143554A TWI564554B (en) | 2014-12-12 | 2014-12-12 | A real-time xxy platform fault diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI564554B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113411019A (en) * | 2021-06-23 | 2021-09-17 | 苏州电器科学研究院股份有限公司 | Motor master-slave chaotic synchronization control method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1125121B1 (en) * | 1998-10-28 | 2007-12-12 | Covaris, Inc. | Apparatus and methods for controlling sonic treatment |
| TW201319539A (en) * | 2011-11-07 | 2013-05-16 | Univ Nat Changhua Education | Measurement system for diagnosing ball screw preload loss by vibration signal and voiceprint signal |
-
2014
- 2014-12-12 TW TW103143554A patent/TWI564554B/en not_active IP Right Cessation
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1125121B1 (en) * | 1998-10-28 | 2007-12-12 | Covaris, Inc. | Apparatus and methods for controlling sonic treatment |
| TW201319539A (en) * | 2011-11-07 | 2013-05-16 | Univ Nat Changhua Education | Measurement system for diagnosing ball screw preload loss by vibration signal and voiceprint signal |
Also Published As
| Publication number | Publication date |
|---|---|
| TW201621292A (en) | 2016-06-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cui et al. | HVSRMS localization formula and localization law: Localization diagnosis of a ball bearing outer ring fault | |
| CN106104496B (en) | The abnormality detection not being subjected to supervision for arbitrary sequence | |
| CN106447040B (en) | Mechanical equipment health state evaluation method based on Heterogeneous Multi-Sensor Data fusion | |
| US20170307360A1 (en) | Status determination device and status determination method | |
| JP6882397B2 (en) | Abnormal noise detection system, equipment, method and program | |
| CN107209512A (en) | Acoustics or the method, system and the computer program that learn the stage of vibration analysis for machine | |
| CN106662504A (en) | Rotating machine abnormality detection device and method and rotating machine | |
| JP2012018066A (en) | Device for inspecting abnormality | |
| Esarey | Fractionally integrated data and the autodistributed lag model: Results from a simulation study | |
| TWI564554B (en) | A real-time xxy platform fault diagnosis method | |
| JP2020144096A (en) | Condition monitoring/diagnosis method of object by multi-directional signal merging and diagnostic system | |
| CN106949925A (en) | It is a kind of to be used for engine, the online visible detection method of gearbox surface coating | |
| Ribeiro et al. | Mechanical fault detection in electric motors measured by a digital signal processing device in an optical mouse | |
| Ma et al. | Two-stage damage identification based on modal strain energy and revised particle swarm optimization | |
| CN106682308B (en) | Load analysis method for load position of coal mining machine | |
| WO2020162425A1 (en) | Analysis device, analysis method, and program | |
| TW201633025A (en) | Diagnostic method for malfunction mode of machine tool main shaft and system thereof | |
| JP7433648B2 (en) | Anomaly detection method | |
| JP2014032122A5 (en) | ||
| WO2021131158A1 (en) | Magnetic detection system, magnetic signal waveform pattern classification method, and waveform pattern distribution creation method for magnetic detection system | |
| JP6746348B2 (en) | Method and device for identifying building stiffness of a building | |
| Trendafoilova et al. | An investigation on vibration-based damage detection in circular plates | |
| Martinod et al. | Estimation of combustion engine technical state by multidimensional analysis using SVD method | |
| CN106548137B (en) | Two degree freedom system Identification of Structural Parameters method based on vibration response signal | |
| KR102313723B1 (en) | Fault Diagnosis Device and Method of Rotating Equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MM4A | Annulment or lapse of patent due to non-payment of fees |