201116361 六、發明說明： - 【發明所屬之技術領域】 • 本發明是有關於一種工具機損耗監測裝置以及 法，更特別的是關於一種利用數據轉換處理震動訊。 以監測工具機刀具磨損狀態之監測裝置以及方法I儿’ 【先前技術】 φ 在機械加工中工具機扮演一個重要的角色，工具機 之係由複數個零件所組成。其中零件中又有磨耗性^件 如刀具等，當工具機在運轉過程中，隨著操作條件不同， 磨耗性零件之磨耗速度亦隨之不同。早期工業界對設°備 之維護，係採用定期保養或當設備出現故障時才被=維 修。 、 近年來企業為提升本身競爭力、追求高效率以及$ 品質之生產技術，全面將設備系統化以及自動化，因^ • 各系統間之關聯性也越來越密切，一旦其中之一設備發 -生故障，所引發之經濟損失也非常可觀。此外，若設備 在不斷運轉之狀態下，無法由外觀直接判斷設備之零件 ，耗知度。因此，主動式工具機磨損監測，可提高設備 官理效能，進而避免設備之突發故障所造成製程上之損 失，且藉由監測狀態可更了解設備之狀態，在需要時進 行檢修。 θ 工具機刀具磨耗監測可分為直接測量以及非直接測 夏。其中直接監測可包含以影像方式以及光學方式，其 201116361 優點為可準確獲得工具機零件之正確磨耗之幾何改變。 然而，由於工具機與加工物連續接觸運轉，因此直接測 量係非常難以執行。 傳統之監測裝置乃利用傅立葉轉換將震動訊號作處 理與資料分析。傅立葉頻譜分析為頻域分析法中最具有 代表性的一種，其能提供便捷之方法來分析資料在頻率 域（frequency domain)的能力量分佈，若訊號是穩態 (stationary)分佈且線性時間序列，即能有效透過頻譜轉 換呈現訊號之特性，然，在機械故障之領域中，所擷取 到之訊號幾乎都是非線性非穩態的時變訊號。對於非穩 態（nonstationary)與非線性（nonlinear)之資料，傅立葉轉 換（Fourier Transformation, FFT)仍顯不足。 因此，需要一種工具機監測裝置以及方法，以非直 接測量方式測量震動訊號（即測量加速度訊號），並藉由 希爾伯特-黃轉換（Hilbert-Huang Transformation, HHT)對 非線性非穩態的時變之震動訊號作處理，並在操作中即 時監測工具機狀態，以防止工具機零件過度磨耗。 【發明内容】 有鑑於上述習知技藝之問題，本發明之其中一目的 就是在提供一種工具機監測裝置以及方法，在操作中即 時監測工具機狀態，藉由希爾伯特-黃轉換針對非穩態與 非線性訊號做訊號處理，進而防止工具機零件過度磨 耗，以解決傳統傅立葉轉換對於非穩態與非線性訊號無 201116361 法處理之問題。 根據本發明之一目的’提出一種工具機監測裝置， 其包含：一工具機、一感測裝置、一訊號處理裝置以及 一顯示裝置。感測裝置，係量測工具機之一震動模式， 以對應產生一震動訊號。訊號處理裝置，用以擷取震動 訊號’並將震動訊號以數據轉換處理以成為一轉換訊 號。顯示裝置，係連接訊號處理裝置接收並顯示轉換訊 號。201116361 VI. Description of the invention: - [Technical field to which the invention pertains] The present invention relates to a machine tool loss monitoring device and method, and more particularly to a method for processing vibration signals using data conversion. Monitoring device and method for monitoring the tool wear state of the machine tool ‘ prior art φ The machine tool plays an important role in machining, and the machine tool is composed of a plurality of parts. Among them, there are wear parts such as tools, etc. When the machine tool is in operation, the wear speed of the wear parts will be different with different operating conditions. In the early days of industry, maintenance was carried out on a regular basis or in the event of equipment failure. In recent years, companies have systematically and automated equipment to improve their competitiveness, pursuit of high efficiency and quality of production technology, because the correlation between systems is becoming more and more close, once one of the devices is issued - The economic losses caused by failures are also very impressive. In addition, if the equipment is in continuous operation, the parts of the equipment cannot be directly judged by the appearance. Therefore, the active tool machine wear monitoring can improve the equipment's official performance, thus avoiding the process loss caused by the sudden failure of the equipment, and by monitoring the state, the state of the equipment can be better understood and the maintenance can be performed when needed. θ Tool tool tool wear monitoring can be divided into direct measurement and indirect summer measurement. Direct monitoring can include both image and optical methods, and its 201116361 has the advantage of accurately obtaining the geometrical changes in the correct wear of the machine tool parts. However, direct measurement is very difficult to perform because the machine tool is in continuous contact with the workpiece. The traditional monitoring device uses Fourier transform to process the vibration signal and analyze the data. Fourier spectrum analysis is the most representative one of the frequency domain analysis methods. It can provide a convenient method to analyze the distribution of data in the frequency domain. If the signal is a stationary distribution and a linear time series. That is, the characteristics of the signal can be effectively transmitted through the spectrum conversion. However, in the field of mechanical failure, the signals captured are almost all non-linear unsteady time-varying signals. For nonstationary and nonlinear data, Fourier Transformation (FFT) is still insufficient. Therefore, there is a need for a machine tool monitoring apparatus and method for measuring a vibration signal (i.e., measuring an acceleration signal) in an indirect measurement manner and for non-linear non-steady state by Hilbert-Huang Transformation (HHT). The time-varying vibration signal is processed and the state of the machine tool is monitored in real time to prevent excessive wear of the machine tool parts. SUMMARY OF THE INVENTION In view of the above-mentioned problems of the prior art, one of the objects of the present invention is to provide a machine tool monitoring apparatus and method for monitoring the state of the machine tool in real time during operation, and aiming at instability by Hilbert-yellow conversion The state and the nonlinear signal are processed by the signal to prevent the tool machine parts from being excessively worn, so as to solve the problem that the conventional Fourier transform has no 201116361 method for the non-steady state and the nonlinear signal. According to one aspect of the present invention, a machine tool monitoring apparatus is provided, comprising: a machine tool, a sensing device, a signal processing device, and a display device. The sensing device measures a vibration mode of the machine tool to generate a vibration signal correspondingly. The signal processing device is configured to capture the vibration signal ’ and convert the vibration signal into a data conversion process to become a conversion signal. The display device is connected to the signal processing device to receive and display the conversion signal.
其中工具機在一第一次運轉時，震動模式為一第一 震動模式，以及工具機在一第二次運轉時，震動模式為 一第二震動模式，感測裝置係對應第一震動模式產生一 第一震動訊號，以及對應第二震動模式產生」第二震動 訊號，經由訊號處理裝置擷取第一震動訊號以及=二震 動訊號，並績據轉換處理轉換成為第—轉換訊號以及 ::轉換訊號，由顯示裝置接收並顯示第一轉換訊號以 及第二轉換訊號。 根據本發明之一目的’又提出一種 =，包含：啟動-工具機運轉。感測器根據工具」 ^次運轉對應產生之—第—震_心量測彳^ ^號雷=據，在一第二:欠運轉時對應產 ^第一震動模式以罝測得一第二震動訊號。以一訊 處理裝置_第-震動訊號以及第二 ° =法轉換一第一轉換訊號以及一第二轉:訊號: 乂一轉換訊號以及第二轉換訊號以決定是否須發出 201116361 告訊5虎至警告裝置。 承上所述，依本發明之具有以希爾伯特-黃轉換 (HHT)將震動訊號作處理之監測裝置以及監測方法，其 可具有一或多個下述優點： (1) 此監測裝置以及方法可對非線性、非穩態之震 動訊號作數據轉換轉換，.分辨不同之轉換訊號的差異， 以判斷出工具機刀具之磨損狀態以及磨損位置。 (2) 此監測裝置以及方法可在工具機運轉狀態下， 即時以非直接之方法監測工具機刀具之磨損狀態以及磨 損位置。 【實施方式】 傳統工具機刀具磨損監測裝置以及方法，係以傅立 葉轉換（Fourier Transformation，FFT)對工具機之加速度 訊號做頻域分析，其頻譜無法真實地反映出非線性或非 穩態的特質，然，一般機械的故障訊號皆為非線性、非 穩態的，因此，本發明係採用黃鍔博士所提出之希爾伯 特-黃轉換（Hilbert-Huang Transformation，HHT)理論，其 使用新的時域訊號分析方法，透過經驗模態分解法 (Ensemble Empirical Mode Decomposition，EMD)將資料 變化的内部時間尺度作為能量，從高頻至低頻將其分解 成多個内建模態函數（Intrinsic Mode Function, IMF)分 量。内建模態函數(IMF)之定義為：函數之任何極值數目 201116361 等於跨零點（咖__sing)數目，以及其包絡線(_㈣ 係對稱於零之函數。 此刀解法係有系統解析出IMF的方法稱之為轉移過 程（Shifting Process)，其分別採用局部極大值（i〇cal maxima)以及局部極小值（1〇cal minima)所定義之包絡 線’先找出訊號巾之局部極A值當作上圍的包絡線，然 後利用立方弧線（cubic spline)把它們連接起來，再找 #出訊號中之所有局部極小值，❹立方弧線產生下圍的 包絡線’然後再取極大值包絡線與極小值包絡線的均值 包絡線（mean envelope)，取名為如，而原始訊號x(t) 與均值mi之差即為第一分量h丨： hi=x(t)-mi ⑴ 再第二次轉移過程中，把hi當作原來訊號，然後 hi-mn= hn • (2) 可重複轉移過程k次，直至ij hlk變成一個w，即 hlk ’最後指定它為Cl= hlk。 為使IMF可有不錯之希爾伯特·黃轉換之特性，且確 保IMF分量之振幅及頻率變動都能保有足夠之物理意 義，必須滿足以下條件：（1)極值數目與跨零2 (—crossing)數目之差需;以及⑵上包絡線（由局 部極大值所連接）以及T包絡線（由局部極小值所連接） 之均值在任何點需為零。Cl為第一卿，其可從原來訊 201116361 號中分離ci成n = s- ci，其中差值η為餘數。再將η 當作新訊號，再重複其轉移過程來作處理。最終，當餘 數rn&法被分解即停止轉移過程，^無法被分解係指rn 變成單調的函數（monotonic function)或函數無法再分 離出具有IMF特性的訊號。最後可得： x(t) = tcj^n (3) Μ 此外，另一步驟對分解的IMF作希爾伯特-黃轉換 (HHT)，每一分量經希爾伯特-黃轉換後為yi: yi=i]^r (4) π -ί ί~τ 經過希爾伯特-黃轉換，其解析訊號z(t)可表示為： z(t)=x(t)+iy(t)=a(t)e，m (5) a(t)^^x2 +y2 以及 0(〇 = tan_1(Z) (6,7)In the first operation of the machine tool, the vibration mode is a first vibration mode, and when the machine tool is in a second operation, the vibration mode is a second vibration mode, and the sensing device is generated corresponding to the first vibration mode. a first vibration signal, and a second vibration signal corresponding to the second vibration mode, the first vibration signal and the second vibration signal are captured by the signal processing device, and the conversion processing is converted into the first conversion signal and the: conversion The signal is received by the display device and displays the first conversion signal and the second conversion signal. According to one of the objects of the invention, a = is also proposed, comprising: start-tool machine operation. The sensor is generated according to the tool "^ times the corresponding operation - the first - shock _ heart measurement 彳 ^ ^ mine = data, in a second: under operation corresponding to the first vibration mode to detect a second Vibration signal. Converting a first conversion signal and a second rotation by a processing device_first-vibration signal and a second °= method: a signal: a first conversion signal and a second conversion signal to determine whether a 201116361 message is required to be issued. Warning device. According to the present invention, there is provided a monitoring device and a monitoring method for processing a vibration signal by Hilbert-Huang Transform (HHT), which may have one or more of the following advantages: (1) The monitoring device And the method can perform data conversion conversion on the nonlinear, non-steady state vibration signal, and distinguish the difference of different conversion signals to determine the wear state and the wear position of the tool tool. (2) The monitoring device and method can monitor the wear state and wear position of the machine tool tool in an indirect way when the machine tool is running. [Embodiment] The conventional tool machine tool wear monitoring device and method are Fourier transform (FFT) for frequency domain analysis of the acceleration signal of the machine tool, and the spectrum cannot truly reflect the nonlinear or non-steady state characteristics. However, the general mechanical fault signals are both non-linear and non-steady. Therefore, the present invention adopts the Hilbert-Huang Transformation (HHT) theory proposed by Dr. Huang Wei, which uses new The time domain signal analysis method uses the internal time scale of data change as energy by Ensemble Empirical Mode Decomposition (EMD), and decomposes it into multiple internal modeling state functions from high frequency to low frequency (Intrinsic Mode) Function, IMF) component. The internal modeling state function (IMF) is defined as: the maximum number of functions of the function 201116361 is equal to the number of crossing zeros (coffee __sing), and its envelope (_ (four) is a function of symmetry to zero. This knife solution is systematically resolved The IMF method is called the Shifting Process, which uses the local maximum (i〇cal maxima) and the local minimum (1〇cal minima) to define the envelope of the signal. The value is taken as the envelope of the upper circumference, and then they are connected by cubic spline, and then all local minimum values in the #out signal are found, and the cubic arc produces the envelope of the lower circle' and then the maximum envelope is taken. The mean envelope of the line and the minimum envelope is named as, and the difference between the original signal x(t) and the mean mi is the first component h丨: hi=x(t)-mi (1) In the second transfer process, hi is treated as the original signal, then hi-mn=hn • (2) The transfer process can be repeated k times until ij hlk becomes a w, ie hlk 'finally specifies it as Cl= hlk. Make the IMF have a good Hilbert-Yellow conversion feature And to ensure that the amplitude and frequency variation of the IMF component can maintain sufficient physical meaning, the following conditions must be met: (1) the difference between the number of extreme values and the number of crossings (crossing); and (2) the upper envelope (by local maximization) The value of the connection) and the T envelope (connected by the local minimum value) must be zero at any point. Cl is the first Qing, which can separate ci from the original 201116361 into n = s- ci, where the difference The value η is the remainder. Then η is treated as a new signal, and then the transfer process is repeated for processing. Finally, when the remainder rn& method is decomposed, the transfer process is stopped, and ^ cannot be decomposed to mean rn becomes a monotonic function (monotonic function) Or the function can no longer separate the signal with IMF characteristics. Finally, we can get: x(t) = tcj^n (3) Μ In addition, another step is to perform Hilbert-Huang transform (HHT) on the decomposed IMF. Each component is converted to Hilbert-Yellow by yi: yi=i]^r (4) π -ί ί~τ After Hilbert-Yellow conversion, the analytic signal z(t) can be expressed as: z(t)=x(t)+iy(t)=a(t)e,m (5) a(t)^^x2 +y2 and 0(〇= tan_1(Z) (6,7)
X 其中a(t)為順時振幅；（9為相位函數 瞬時頻率学 (8) at 每一分量作希爾伯特-黃轉換後，原始訊號可表示為實部 R以以下之形式： X (t)=R { 2 (〇 exp[/ |ω; (t)dt\ (9) 7=1 以希爾伯特頻譜（Hilbert Spectrum)定義，可定義出邊際 頻譜(Marginal Spectrum)為： 201116361 (10) 即表達出整 邊際頻譜提供對於每_率的總振幅量測 個時間長度所累積的振幅（或能量）。、 特性對都能具有良好之希爾伯特_黃轉換之 寺隹。由於刀離之過程僅依據訊局部的時間尺 行，因此，能處理不同特性之訊號，如平穩、非ς穩訊X where a(t) is the clockwise amplitude; (9 is the phase function instantaneous frequency (8) at each component after Hilbert-yellow conversion, the original signal can be expressed as the real part R in the following form: X (t)=R { 2 (〇exp[/ |ω; (t)dt\ (9) 7=1 Defined by Hilbert Spectrum, Marginal Spectrum can be defined as: 201116361 (10) That is, the full marginal spectrum is expressed to provide the amplitude (or energy) accumulated for the total amplitude of each _ rate measured for a certain length of time. The characteristic pair can have a good Hilbert_Yellow conversion temple. Since the process of knife separation is based only on the local time scale of the signal, it can handle signals with different characteristics, such as smooth and non-stable signals.
號與線性、雜性訊號。經由希爾伯特_黃轉換\ 得瞬時訊號鮮及振幅。最後可表示成振幅.頻^時間 之三維分佈圖，稱之為希爾伯特時頻譜（Hilb如 Spectrum) ° 請參閱第1圖，其係為本發明之工具機監測裝置之示 意圖。圖中，工具機監測裝置100包含工具機1〇2、感 測裝置104、訊號處理裝置106以及顯示裝置1〇8。感測 裝置104係用以量測工具機102之震動模式，以對應產 生震動訊號。訊號處理裝置1〇6係用以掏取該震動訊 说，並將震動訊號作數據轉換以成為轉換訊號，在本實 施例中，係以希爾伯特-黃轉換（Hilbert_Huang Transformation，HHT)進行說明，但本發明於實際實施 時，並不限於此，舉凡能將非線性非穩態的時變之震動 訊號’轉換成可分析變化的訊號者，皆屬本發明所稱之 數據轉換。顯示裝置108係電連接訊號處理裝置1〇6， 用以接收並顯示轉換訊號。 201116361 感測裝置104係三維加速規（acceler〇meter)或三維 應力規（strain guage)。感測裝置1〇4可分別量測工具機 102之三維度方向之加速度值。 工具機102之全新刀具在第一次運轉時，其震賴式為 第一震動模式。又’工具機102之刀具在第二次運轉時，震 動模式為第二震動模式。其中第二運轉模式係指除工具機 1〇2具有全新刀具之第-次運轉模式外之運轉模式。感剛巢 置104係對應第-震動模式產生一第一震動訊號，以及對應 第二震動模式產生一第二震動訊號。第一震動訊號以及第 二震動訊號經由訊號處理裝置106擷取，並以希爾伯特、 黃轉換（HHT)成為第一轉換訊號以及第二轉換訊號。第 一轉換訊號以及第二轉換訊號係由顯示裝置1〇8接收並 顯示。 處理單元1064係將第二轉換訊號與第一轉換訊鱿 比較以得一比較結果，並根據比較結果決定是否發出〜 警告訊號。 希爾伯特-黃轉換（HHT)係以經驗模態分解法 (empirical mode decomposition，EMD)將該震動訊號分解 成複數個IMF分量，並經由希爾伯特-黃轉換，求得瞬時 訊號頻率及振幅。最後可表示成振幅-頻率-時間之三維 分佈圖，稱之為希爾伯特時頻譜（Hilbert Spectrum)。 於另一實施例中，訊號處理裝置106包含類比數仅 201116361 轉換單元1062、處理單元1064以及儲存單元1066。類 比數位轉換單元1062係接收並轉換感測裝置104類比之 震動訊號以成為數位之震動訊號。處理單元1064係接收 數位之震動訊號，並經由希爾伯特-黃轉換（HHT)以成為 轉換訊號。儲存單元1066係儲存轉換訊號。 於又一實施例中，工具機監測裝置100更包含警告裝 置110,係電連接於處理單元1064。當工具機102之處 φ 理單元1064將第二轉換訊號與儲存於儲存單元中之第 一轉換訊號作比較時，由於第一轉換訊號係為具有全新 刀具之工具機110所測得之訊號，因此，使用者可由第 一轉換訊號以及第二轉換訊號，表示於希爾伯特時頻譜 上之比較結果，以得知目前刀具之磨損狀態，並進一步 分析刀具磨損之位置。 使用者可設定一磨耗臨界值，當處理單元1064比較 φ 第二轉換訊號表示於該希爾伯特時頻譜之頻譜位置超過 磨耗臨界值時，處理單元1064發出一警告訊號至警告裝 置110,並藉由警告裝置110發出警告。 於又一實施例中，工具機監測裝置110更包含電腦系 統112，其係電連接於處理單元1064。電腦系統112接 收轉換訊號，並根據轉換訊號對工具機102作監測。需 說明的是，電腦系統112可藉由一網路，或一無線傳輸 方式接收轉換訊號。 11 201116361 請參閱第2圖，其係為本發明之工具機監測方法之 一較佳實施例流程圖。工具機刀具磨耗監測方法包含： 步驟S202係啟動工具機運轉。步驟S204係感測器根據 工具機具有全新刀具之第一次運轉對應產生之第一震動 模式，以量測得第一震動訊號，以及根據工具機在第二 次運轉時對應產生之第二震動模式以量測得第二震動訊 號。以訊號處理裝置擷取第一震動訊號以及第二震動訊 號，並以希爾伯特-黃轉換 （Hilbert-Huang Transformation, HHT)轉換為第一轉換訊號以及第二轉 換訊號。步驟S 210係比較第一轉換訊號以及第二轉換訊 號以決定是否須發出一警告訊號。 其中，希爾伯特-黃轉換(HHT)之步驟更包含步驟 S206 經驗模態分解法（empirical mode decomposition, EMD)將該震動訊號分解成複數個内建模態函數 (Intrinsic Mode Function, IMF)分量。以及步驟 S208，係 將IMF分量並經由希爾伯特-黃轉換作信號處理，進而可 求得瞬時訊號頻率及振幅。最後可表示成振幅-頻率-時 間之三維分佈圖，稱之為希爾伯特時頻譜（Hilbert Spectrum) ° 當工具機之處理單元將第二轉換訊號與儲存於儲 存單元中之第一轉換訊號作比較時，由於第一轉換訊號 係為具有全新刀具之工具機所測得之訊號，因此，使用 者可由第一轉換訊號以及第二轉換訊號，表示於希爾伯 12 201116361 特時頻譜上之比較結果，以得知目前刀具之磨損狀態， 並進一步分析刀具磨損之位置。 使用者可設定一磨耗臨界值，當處理單元比較第二 轉換訊號表示於該希爾伯特時頻譜之頻譜位置超過磨耗 臨界值時，處理單元發出一警告訊號至警告裝置，並於 步驟S212藉由警告裝置發出警告。 於一實施例中，工具機監測方法更包含S214步驟， 傳送第一轉換訊號以及第二轉換訊號至顯示裝置。 訊號處理裝置係包含類比數位轉換單元、處理單元 以及儲存單元。類比數位轉換單元係接收並轉換該感測 裝置類比之第一震動訊號以及第二震動訊號以成為數位 之第一震動訊號以及第二震動訊號。處理單元係接收數 位之第一震動訊號以及第二震動訊號，並經由希爾伯特-黃轉換（HHT)以成為第一轉換訊號以及第二轉換訊號， 以及儲存單元係儲存第一轉換訊號以及第二轉換訊號。 於另一實施例，工具機監測方法更包含步驟S216 係傳送第一轉換訊號以及第二轉換訊號至電腦系統。需 說明的是，電腦系統可藉由網路，或無線傳輸方式接收 轉換訊號。 感測裝置係三維加速規（accelerometer)或一三維應 力規（strain guage)，用以分別量測工具機之三維度方向 之加速度值。 於一實施例中，請參閱第3圖，其係本發明之工具 13 201116361 機監測裝置之感測器所測得之加速度_時間圖。由第3圖 所示，難以藉由此圖分辨出磨損刀具之加速度大小大於 全新刀具之加速度大小。由於故障訊號係非線性、非穩 態之訊號，因此，藉由經驗模態分解法（empiHcal decomposition，EMD)分解成IMF分量，如第4圖所示。 最後，由公式（10)可繪出X軸方向加速度之邊際頻譜 圖，如第5圖所示。以及繪出2軸方向加速度之邊際頻 譜圖，如第6圖所示。由第5圖可看出，以全新刀具之 工具機其峰值為129Ηζ作為基準，相對於磨損刀具隻工 具機其峰值為117. 5Hz。因此，由邊際頻譜圖中，可明 顯分辨出峰值之不同，進而可及時監測卫具機刀具之磨 耗狀態。 以上所述僅為舉例性，而非為限制性者。任何未脫 離本發明之精神與料，”其進行之等效修改或變 更，均應包含於後附之申請專利範圍中。 【圖式簡單說明】 第1圖係為本發明之工具機監測裝置之一較佳實施例之 示意圓； 第2圖係、為本發明之工具機監測方法之之—較佳實施例 流程圖； 201116361 第3圖係本發明之工具機監測裝置之另一較佳實施例 之感測器所測得之加速度-時間圖； 第4圖係本發明之工具機監測裝置之另一較佳實施例 以EMD將故障訊號分解成IMF分量圖； 第5圖係本發明之工具機監測裝置之另一較佳實施例 餘X軸方向之加速度之邊際頻譜圖；以及 第6圖係本發明之工具機監測裝置之另一較佳實施例 • 餘Z軸方向之加速度之邊際頻譜圖。 【主要元件符號說明】 100 :工具機監測裝置； 102 :工具機； 104 :感測裝置； 106 :訊號處理裝置；Number and linear, heterozygous signal. Transient signal and amplitude via Hilbert_Yellow conversion. Finally, it can be expressed as a three-dimensional map of amplitude, frequency, time, called Hilbert time spectrum (Hilb, Spectrum). Please refer to Fig. 1, which is a schematic diagram of the machine tool monitoring device of the present invention. In the figure, the machine tool monitoring device 100 includes a machine tool 1, a sensing device 104, a signal processing device 106, and a display device 1〇8. The sensing device 104 is configured to measure the vibration mode of the power tool 102 to generate a vibration signal. The signal processing device 1〇6 is configured to capture the vibration signal and convert the vibration signal into data to be a conversion signal. In this embodiment, the Hilbert_Huang Transformation (HHT) is used. It should be noted that, in the actual implementation of the present invention, it is not limited thereto, and any one that can convert a non-linear unsteady time-varying vibration signal into a signal that can be analyzed and changed is referred to as data conversion in the present invention. The display device 108 is electrically connected to the signal processing device 1〇6 for receiving and displaying the conversion signal. The 201116361 sensing device 104 is a three-dimensional accelerometer or a three-dimensional strain gauge. The sensing device 1〇4 can measure the acceleration values of the three-dimensional direction of the power tool 102, respectively. When the new tool of the machine tool 102 is operated for the first time, the shock mode is the first vibration mode. Further, in the second operation of the tool of the machine tool 102, the vibration mode is the second vibration mode. The second mode of operation refers to an operation mode other than the first-time operation mode in which the machine tool 1〇2 has a new tool. The sense nest 104 series generates a first vibration signal corresponding to the first vibration mode, and generates a second vibration signal corresponding to the second vibration mode. The first vibration signal and the second vibration signal are captured by the signal processing device 106, and converted into a first conversion signal and a second conversion signal by Hilbert and Yellow (HHT). The first switching signal and the second switching signal are received and displayed by the display device 1A8. The processing unit 1064 compares the second conversion signal with the first conversion signal to obtain a comparison result, and determines whether to issue a ~warning signal according to the comparison result. Hilbert-Huang transform (HHT) decomposes the vibration signal into a plurality of IMF components by empirical mode decomposition (EMD) and obtains the instantaneous signal frequency via Hilbert-Huang transform. And amplitude. Finally, it can be expressed as a three-dimensional distribution of amplitude-frequency-time, called Hilbert Spectrum. In another embodiment, the signal processing device 106 includes an analogy only 201116361 conversion unit 1062, a processing unit 1064, and a storage unit 1066. The analog digital conversion unit 1062 receives and converts the analog signal of the sensing device 104 to become a digital vibration signal. The processing unit 1064 receives the digital vibration signal and converts it into a switching signal via Hilbert-Huang transform (HHT). The storage unit 1066 stores the conversion signal. In yet another embodiment, the power tool monitoring device 100 further includes a warning device 110 electrically coupled to the processing unit 1064. When the tool unit 102 compares the second conversion signal with the first conversion signal stored in the storage unit, since the first conversion signal is a signal measured by the machine tool 110 having the new tool, Therefore, the user can display the comparison result of the Hilbert time spectrum by the first conversion signal and the second conversion signal to know the current wear state of the tool and further analyze the position of the tool wear. The user can set a wear threshold. When the processing unit 1064 compares the φ second conversion signal to indicate that the spectral position of the spectrum of the Hilbert exceeds the wear threshold, the processing unit 1064 sends a warning signal to the warning device 110, and A warning is issued by the warning device 110. In yet another embodiment, the power tool monitoring device 110 further includes a computer system 112 that is electrically coupled to the processing unit 1064. The computer system 112 receives the conversion signal and monitors the machine tool 102 based on the conversion signal. It should be noted that the computer system 112 can receive the conversion signal by means of a network or a wireless transmission. 11 201116361 Please refer to FIG. 2, which is a flow chart of a preferred embodiment of the machine tool monitoring method of the present invention. The tool machine tool wear monitoring method comprises: Step S202 is to start the machine tool operation. Step S204 is based on the first vibration mode generated by the first operation of the machine tool having the new tool, to measure the first vibration signal, and according to the second vibration generated by the machine tool during the second operation. The mode measures the second vibration signal. The first vibration signal and the second vibration signal are captured by the signal processing device, and converted into a first conversion signal and a second conversion signal by Hilbert-Huang Transformation (HHT). Step S210 compares the first conversion signal and the second conversion signal to determine whether a warning signal is to be issued. The Hilbert-Yellow Transformation (HHT) step further includes the step S206 empirical mode decomposition (EMD) to decompose the vibration signal into a plurality of Intrinsic Mode Function (IMF). Component. And in step S208, the IMF component is converted into a signal processing via Hilbert-Yellow, and the instantaneous signal frequency and amplitude are obtained. Finally, it can be expressed as a three-dimensional distribution of amplitude-frequency-time, called Hilbert Spectrum. When the processing unit of the machine tool uses the second conversion signal and the first conversion signal stored in the storage unit. For comparison, since the first conversion signal is a signal measured by a machine tool having a new tool, the user can display the first conversion signal and the second conversion signal on the Hilbert 12 201116361 special time spectrum. Compare the results to find out the current wear state of the tool and further analyze the position of the tool wear. The user can set a threshold of wear. When the processing unit compares the spectrum position of the spectrum when the second conversion signal is indicated by the Hilbert, the processing unit sends a warning signal to the warning device, and borrows from the alarm device in step S212. A warning is issued by the warning device. In an embodiment, the machine tool monitoring method further includes the step S214 of transmitting the first conversion signal and the second conversion signal to the display device. The signal processing device includes an analog digital conversion unit, a processing unit, and a storage unit. The analog digital conversion unit receives and converts the first vibration signal and the second vibration signal analogous to the sensing device to become the first vibration signal and the second vibration signal of the digital position. The processing unit receives the first vibration signal and the second vibration signal of the digits, and is converted into a first conversion signal and a second conversion signal by a Hilbert-Huang transform (HHT), and the storage unit stores the first conversion signal and The second conversion signal. In another embodiment, the machine tool monitoring method further includes the step S216 of transmitting the first conversion signal and the second conversion signal to the computer system. It should be noted that the computer system can receive the conversion signal through the network or wireless transmission. The sensing device is a three-dimensional accelerometer or a three-dimensional strain gauge for measuring the acceleration value of the three-dimensional direction of the machine tool. In an embodiment, please refer to FIG. 3, which is an acceleration_time diagram measured by a sensor of the tool 13 201116361 machine monitoring device of the present invention. As shown in Figure 3, it is difficult to distinguish from the figure that the acceleration of the worn tool is greater than the acceleration of the new tool. Since the fault signal is a nonlinear, unsteady signal, it is decomposed into IMF components by empirical mode decomposition (EMD), as shown in Fig. 4. Finally, the marginal spectrum of the X-axis acceleration can be plotted by equation (10), as shown in Figure 5. And plot the marginal spectrum of the acceleration in the 2-axis direction, as shown in Figure 6. It can be seen from Fig. 5 that the peak value of the machine tool with the new tool is 129 Ηζ, and the peak value of the tool is 117. 5 Hz. Therefore, from the marginal spectrum diagram, the difference in peak value can be clearly distinguished, and the wear state of the guard tool can be monitored in time. The above is intended to be illustrative only and not limiting. Any equivalent modifications or alterations of the present invention should be included in the scope of the appended claims. [FIG. 1] FIG. 1 is a machine tool monitoring device of the present invention. A schematic circle of a preferred embodiment; FIG. 2 is a flow chart of a preferred embodiment of the machine tool monitoring method of the present invention; 201116361 FIG. 3 is another preferred embodiment of the machine tool monitoring device of the present invention. The acceleration-time diagram measured by the sensor of the embodiment; FIG. 4 is another preferred embodiment of the machine tool monitoring apparatus of the present invention, which decomposes the fault signal into an IMF component map by EMD; Another preferred embodiment of the machine tool monitoring device is a marginal spectrogram of the acceleration in the X-axis direction; and FIG. 6 is another preferred embodiment of the machine tool monitoring device of the present invention. Marginal spectrum diagram [Description of main component symbols] 100: Machine tool monitoring device; 102: Machine tool; 104: Sensing device; 106: Signal processing device;
1062 :類比數位轉換單元； 1064 :處理單元； 1066 :儲存單元； 108 :顯示裝置； 110 :警告裝置； 112 :電腦系統；以及 S202〜S216 :步驟。 151062: analog-to-digital conversion unit; 1064: processing unit; 1066: storage unit; 108: display device; 110: warning device; 112: computer system; and S202~S216: steps. 15