200905208 九、發明說明： 【發明所屬之技術領域】 本發明涉及一種信號分析系統及方法， 關]士 號轉態區間非單調性的分析系統及方法。 ” ° 【先前技術】 兵百鬲階圖形介面的測量儀器在信號測量中被廣泛 應用’如心電圖儀、頻講分析儀、示波器等。其中最常用 的儀是不波器，用於觀察信號的波形、頻率、幅产等。 仗分析手段來說，當示波器橫轴表示時間，縱軸爲電壓幅 度，波形曲線表示電壓隨時間變化的執迹時，我們稱之爲 時域測量；當城H橫座標錢率，縱軸爲功率幅，波= 曲線表示不同頻率點上功率幅度的分料，我們稱之爲頻 域分析。 一八 如圖4中所示’是從用於測量1C (integrate circ則晶 片電壓信號的示波ϋ上㈣的—段反映⑽發生轉態的波 形曲線，觀圖4可知該波形曲線整體呈現出非單調遞增/遞 減的特性：鑛齒形曲線逐步上升至某—點（假設該點爲ρι) 後呈現下降趨勢，下降至某—最低點（假設該點ρ2)後又 轉而王現上升趨勢。在信號分析過程巾，確定轉態區間内 的峰值點（如所述!^點或打點），可以應用在大部份ic (Integrate Circuit)晶片彼此間以類比方式傳輪訊號時， 1C晶片接收端判定〇與1的基準電壓（VMgh & vi〇w)附近 是否有波動，藉以確定1C晶片是否違反設計規範，以及用 於判斷1C晶片的參考電壓（Vref)附近是否有非單調性之小 7 200905208 波存在以對信號進行抖動分析（Jitter Analysis)有重大影 . 響。同時亦可以依據峰值點確定分段區間，對此分段區間 進行傅立葉（FFT)轉換變時域分析爲頻域分析，^二 波動高頻信號在轉態區間内之主頻率與頻域上之能量分 佈0 目前，在利用高階圖形介面的測量儀器測量信號的過 桎中，對於轉態區間的非單調性與峰值點，常常依賴於人 眼視覺辨別、判定。 、 然而藉由人工測量轉態區間的非單調性及峰值點存 在許多問題，諸如：i•大量的量測資料中可能只有很少數 的特定波形具備非單調性；2.對峰值點定位不精確可能導 致對1C晶#基準電壓是否有波㈣斷錯誤，以及導致進行 FFT轉換時的分段區間不精確。 【發明内容】 鑒於以上内容，有必要提供一種信號轉態區間非單調 《性的分析系統及方法，可以從大量的信號測量資料中準確 ^找出少數具備非單雛的波形#料，並對該非單調性轉 態區間内的峰值發生點進行精確定位。 ^ 一種號轉態區間非單調性的分析系統，該系統包括 ^虎測里儀及測量分析裝置。該信號測量儀接收信號發生 裝^發出的信號並以波形曲線你顯示信號的變化情況。該 測量分析裝置包括：資料擷取單元，用於從信號測量儀擷 \波开^曲線y⑷的數值資料，包括信號波形曲線及 信號的引數參數x在一轉態區間内的數值集合及[X,]， 8 200905208 ^自然數;曲線擬合單元，用於根據上述 =數值集合曲線擬合求取波形曲線 了奴， ⑷’4异早几，用於對所求得的最佳逼近 C對r該轉態區間上的自變參數”進行—階與二階 二Γ佳逼近函數〜在該轉態區間上的-階離 值w及二階離散微分值集合；定位單元，用於柃 階離散微分值集合中是否有零值，若有零值，： 二明取佳逼近函數心在該轉態區間上具備非單調性，以該 零值所對應的函數〜曲線上的所述峰值點（r χ ^點’獲取該峰值點的曲率半徑R，截取2倍於料值點 ）曲率半徑R的引數區間U,R、+R]，對該2R 長度區間内的數值㈣進行比較，得到波形曲線吻在該 2R長度㈣㈣峰值位置及鱗值位置的座標。 、-種錢轉態區㈣單調性的分析方法，該方法包括 二下步驟.（A)信號測量儀接收信號發生裝置發出的信號 亚以波形曲線⑽顯示信號的變化情況；（B)從所述n ^量儀擷取轉態區間内信號波形曲線你的數值資料，包^ k號波形曲線灿及信號的引數參數χ在—轉隸間内的數 值市口 [νΜ及[X,]，其中0。。]，”爲自然數；（C)根據上 逑㈣的數值集合_進行曲線擬合求取波形曲線函數 咖的最佳逼近函數FW;(D)對所求得的最佳逼近函數心相 :於f轉態區間上的引數參數x進行-階與二階微：運 异’得到最佳逼近函數4)在該轉態關上的—階離散微分 值集合及二階離散微分值集合，·⑻檢查所述一階離散微 200905208 分值集合巾是否有零值；⑺若有零值，賴取該零值所 對應的函數〇線上的該峰值點座標（^，〜)）及該峰值 :的曲率’（G)以该峰值點（》，^))爲中心截取兩倍於 騎值點曲率半禋R的引數區間卜,_R ”,+r];及⑻比 較該2R長度引數區間内相應的數值㈣，得到信號波形 曲、’泉y㈨在3亥2R長度區間内的峰值位置及該位置的座標。 ^相幸又於白知技術，本發明所提供的信號轉態區間非單 周1"的刀析系統及方法能夠從大量的信號資料中將少數具 備非單則生的的波形資料快速找出來，建立有效樣本波 ^對U虎非單δ周性轉態區間内的峰值發生點進行精確定 位，降低人工作業誤差。 【實施方式】 =方便理解，對文中涉及的部分術語作如下定義： 號‘ I、.，乏彳a k號由低穩態上升至高穩n或高穩態 下降至低穩態。200905208 IX. Description of the Invention: [Technical Field] The present invention relates to a signal analysis system and method, and a system and method for analyzing a non-monotonicity of a transition range. ° [Prior Art] The measuring instruments of the Bentley-level graphical interface are widely used in signal measurement, such as electrocardiographs, frequency analyzers, oscilloscopes, etc. The most commonly used instruments are non-wave devices for observing signals. Waveform, frequency, amplitude, etc. 仗 Analytical means, when the horizontal axis of the oscilloscope represents time, the vertical axis represents voltage amplitude, and the waveform curve represents the trace of voltage change with time, we call it time domain measurement; The horizontal coordinate money rate, the vertical axis is the power amplitude, the wave = curve represents the power amplitude at different frequency points, we call it the frequency domain analysis. One of the eight is shown in Figure 4 'is used to measure 1C (integrate Circ, the oscilloscope of the wafer voltage signal (4) reflects the waveform of the transition (10), and Figure 4 shows that the waveform exhibits a non-monotonically increasing/decreasing characteristic: the ore profile gradually rises to a certain value. The point (assuming the point is ρι) shows a downward trend, and drops to a certain low point (assuming the point ρ2) and then turns to the king's rising trend. In the signal analysis process, the peak point in the transition interval is determined. As described above, the ^^ point or dot) can be applied to the analog voltage of the majority of ic (Integrate Circuit) wafers. The 1C chip receiving terminal determines the reference voltage of 〇 and 1 (VMgh & vi〇w Whether there is fluctuation in the vicinity to determine whether the 1C wafer violates the design specifications, and whether there is a non-monotonicity near the reference voltage (Vref) of the 1C wafer. 7 200905208 Wave exists to analyze the jitter of the signal (Jitter Analysis) It is also possible to determine the segmentation interval based on the peak point. Fourier transform (FFT) transformation time domain analysis is used for frequency domain analysis, and the main frequency of the two-wave high-frequency signal in the transition interval Energy distribution in the frequency domain 0 Currently, in the measurement of signal overshoot using a high-order graphical interface measuring instrument, the non-monotonicity and peak point of the transition interval often depend on human visual discrimination and judgment. There are many problems in manually measuring the non-monotonicity and peak points of the transition interval, such as: i• There may be only a small number of specific waveforms in a large amount of measurement data. Tonality; 2. Inaccurate positioning of the peak point may result in a wave (four) break error for the 1C crystal # reference voltage, and an inaccurate segmentation interval when the FFT conversion is performed. [Inventive content] In view of the above, it is necessary to provide A non-monotonic analysis system and method for signal transition interval, which can accurately identify a small number of non-single-waveform materials from a large number of signal measurement data, and the peak occurrence point in the non-monotonic transition interval Accurate positioning. ^ A non-monotonic analysis system with a range of transitions, including a Huometer and a measurement and analysis device. The signal meter receives the signal generated by the signal and displays the signal as a waveform. Changes. The measurement and analysis device comprises: a data acquisition unit for numerical data from the signal measuring instrument 波\wave opening curve y(4), including a signal waveform and a numerical parameter set of the signal's argument parameter x in a transitional interval and [ X,], 8 200905208 ^Natural number; curve fitting unit, used to obtain the waveform curve according to the above = numerical set curve fitting, (4) '4 different early, for the best approximation obtained C For the self-variable parameter on the transition interval of r, the -order and second-order two-dimensional approximation function are performed - the -th order value w and the second-order discrete differential value set on the transition interval; the locating unit is used for the 柃-order discrete Whether there is a zero value in the set of differential values, if there is a zero value, the second-minded approximation function has non-monotonicity in the transition interval, and the peak point on the function-curve corresponding to the zero value ( r χ ^点' obtains the radius of curvature R of the peak point, intercepts the exponential interval U, R, +R] of the radius of curvature R of the material value point, and compares the value (4) in the 2R length interval to obtain Waveform curve kiss at the 2R length (four) (four) peak position and scale position (b) monotonic analysis method, the method includes two steps. (A) the signal measuring instrument receives the signal from the signal generating device, and the waveform curve (10) shows the change of the signal; (B) From the n ^ meter, take the signal waveform of the transition range. Your numerical data, including the waveform parameter of the k-curve curve and the signal, and the numerical parameters of the signal in the transfer room [νΜ and [X ,], where 0.]," is a natural number; (C) according to the numerical set of the upper (4) _ curve fitting to obtain the best approximation function FW of the waveform curve function coffee; (D) for the obtained The best approximation function: the exponential parameter x on the f-transition interval is -order and second-order micro: the difference is the best approximation function 4) the set of discrete-order differential values and the second-order discretization at the transition state a set of differential values, (8) checking whether the first-order discrete micro-200905208 score set towel has a zero value; (7) if there is a zero value, relying on the peak point coordinate of the function corresponding to the zero value (^, ~ )) and the peak: the curvature '(G) is twice as large as the center point (", ^)) The value of the point curvature is half the radius of the R range of the argument, _R ”, +r]; and (8) compare the corresponding value in the 2R length of the exponent interval (four), get the signal waveform, 'spring y (nine) in the 3H 2R length interval The peak position and the coordinates of the position. ^ Fortunately, in the Baizhi technology, the signal analysis interval and non-single-cycle 1" knife analysis system and method provided by the present invention can have a small number of non-single from a large amount of signal data. The raw waveform data is quickly found out, and the effective sample wave is established to accurately locate the peak occurrence point in the non-single δ circumferential transition interval of the U-hu, and reduce the manual operation error. [Embodiment] = Convenient understanding, relevant to the text Some of the terms are defined as follows: No. 'I,., 彳 ak number rises from low steady state to high steady state n or high steady state decreases to low steady state.
〜圖丨=本發明信號轉態區間非單調性的分析系統較佳 賴圖。該系統包括信號發生器1G、信號測量儀 d里刀析裝置3〇及資料庫4〇。 信號發M 1G可㈣任何産生彳㈣他號的裝置。於 本貫施例中，信號發生哭1〇爲 ^ . 、 信號測量儀2〇具有:^^產^待測八電愿的主機板。 出），且與錢發生㈣電連接，接受__10 ^ 的待測信號，並以波形曲線（也稱作波形圖）表 號的變化。於本實崎，信號測量儀2〇爲一二： 10 200905208 .於監測主機板上電壓的變化情況。 旬曰刀析裳置3〇與信號測量儀20相it，用於對作_ 測置儀20測得的作#咨 T彳°派1 样/、择分r a 貝科進仃自動分析，對信號非單調遞 a/遞減區域的峰值産生點進行精確定位。 貝料庫40用於儲存信號的原始資料以及分析過程中 產生的資料。 如圖2所示，是圖1中測量分析裝置30的功能模組 圖15亥/則里分析裝置30包括：資料操取單元100、曲線擬 合單元110、計算單元120及定位單元13〇。 貧料擷取單元10 0用於從信號測量儀2 0擷取信號波形曲 線y W (如圖4所示）的數值資料，如該波形曲線所對應 的麥數及該參數在一轉態區間上的離散數值集合。於其他 實施例中，資料擷取單元100也可直接擷取資料庫4〇中儲存 的信號的原始數值資料。 曲線擬合單元110用於利用數學方法，如利用多項式 函數、三角函數、對數函數、指數函數，根據所述信號的 離散數值集合進行曲線擬合以求取波形曲線函數y(功的最 佳逼近函數心。本實施例以利用多項式函數進行曲線擬合 爲例說明，該多項式函數的數學通式爲 11 200905208 ⑴=2Λ.Χ <-〇 爲自然數 爲多 項式係數。於本實施例巾，1爲原始信號的自變參數（如 時間T)’記X在該轉態區間上的離散值的集合爲w; y(x)爲 (如 信號測量儀20測得的隨自變參數變化而變化的因變數 電壓V)’記*)在該轉態區間上與X的離散值一對應的離 曲線 t值的集合爲b〇〇]、爲需要求取的最佳逼近函數。 令 擬合單元110求取係數矩障[α,]，如利用求方差的方法： ]2爲最小以得到最~ Figure 丨 = The non-monotonic analysis system of the signal transition interval of the present invention is better. The system includes a signal generator 1G, a knife analyzer 3 in the signal measuring instrument d, and a data bank 4〇. The signal M 1G can be (4) any device that generates the 四 (4) other number. In the present example, the signal is cried 1〇 as ^. The signal measuring instrument 2〇 has: ^^Production ^ The board to be tested. Out), and electrically connected with the money (4), accept the signal to be tested of __10 ^, and change the waveform of the waveform (also called waveform). In this Shisaki, the signal measuring instrument 2 is one or two: 10 200905208. The voltage on the motherboard is monitored. Xun 曰 析 析 置 置 〇 〇 〇 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号 信号The peak generation point of the signal non-monotonic a/decremental region is accurately located. The shell library 40 is used to store the raw data of the signal and the data generated during the analysis. As shown in Fig. 2, it is a functional module of the measurement and analysis device 30 of Fig. 1. Fig. 15 The analysis device 30 includes a data manipulation unit 100, a curve fitting unit 110, a calculation unit 120, and a positioning unit 13A. The lean material extraction unit 10 0 is used to extract the numerical data of the signal waveform curve y W (shown in FIG. 4 ) from the signal measuring instrument 20 , such as the number of wheat corresponding to the waveform curve and the parameter in a transition interval A collection of discrete values on. In other embodiments, the data capture unit 100 can directly retrieve the raw value data of the signal stored in the database. The curve fitting unit 110 is configured to perform a curve fitting according to a discrete numerical set of the signal by using a mathematical method, such as a polynomial function, a trigonometric function, a logarithmic function, an exponential function, to obtain a waveform curve function y (the best approximation of the work) The function of the present embodiment is described by taking a polynomial function for curve fitting as an example. The mathematical formula of the polynomial function is 11 200905208 (1)=2Λ.Χ <-〇 is a natural number is a polynomial coefficient. In this embodiment, 1 is the self-variation parameter of the original signal (such as time T)', and the set of discrete values of X in the transition interval is w; y(x) is (as measured by the signal measuring instrument 20 with the self-variation parameter) The variation factor V of the change V) 'remembers *) The set of off-curve t values corresponding to the discrete values of X in the transition interval is b〇〇], which is the best approximation function to be required. Let the fitting unit 110 obtain the coefficient moment barrier [α,], for example, by using the method of finding the variance: ] 2 is the minimum to get the most
〜與擬合函數y(x)的方差D 佳逼近函數心運算式（如圖5所^曲線^波形曲線 的最佳逼近曲線）。 計算單元120用於對所求得的最佳逼近函數〜相對於 该轉態區間上的自變參數丨1推〜 i」進仃一階與二階微分運算 12 200905208~ The variance D of the fitting function y(x) is better than the function of the heart (as shown in Fig. 5). The calculating unit 120 is configured to push the first-order and second-order differential operations with respect to the obtained optimal approximation function~ relative to the self-variation parameter 丨1 on the transition interval. 12 200905208
dx 在該轉態區間上的一階離散 ;=fl \ Xit-l ~Xi 亦即計算曲線 F, 在轉態區間各離散點的斜率及曲率，得到最佳逼近 r(-〇 欣仉刀值术。[々j及二階離散微 分值集合。 否有用於檢查一階離散微分值集合心中是 否有令值，右有仏=0，則表明最佳逼近函Dx is the first-order dispersion in the transition interval; =fl \ Xit-l ~Xi is the curve F, and the slope and curvature of each discrete point in the transition interval are obtained, and the best approximation r is obtained. [々j and second-order discrete differential value sets. No is used to check whether there is a value in the first-order discrete differential value set, and right 仏=0, indicating the best approximation function.
上具有;，性’定位單元一曲線二】 U ’〜處的曲率仏。若仏>〇，則表明點G， A 曲線〜在轉態區間上的一個波谷（如圖6所示的點(:)2)马 若曲線心<〇’則表明點“、）爲曲線F(舰態區間上 的一個波峰（如圖6所示的點vl)。進一步地，定位單元 13〇以曲線&上的所述峰值點（n)(如點vi或點= 爲中心點，截取2倍於該峰值點（χ，〜)）曲率半徑r(r 爲曲率F(:,)絕對值的倒數）的引數區間h_R，〜+r]，對該 2R長度區間内的數值進行比較，得到波形曲線灿在 該2R長度區間内的峰值位置及該位置的座標，如圖7中 的點 Pm(Xm，y(Xm))，Pn(Xn，y(Xn))。 如圖3所示，是本發明信號轉態區間非單調性的分析 方法較佳實施例的流程圖。首先，在步驟s 1〇，信號測量 儀20接收信號發生器1〇發出的信號，並以波形曲線（記 13 200905208 爲〆X))於彳S號發生為10的—顯示介面上顯示信號的變化 情況。 在步私S12，貧料擷取單元1〇〇從信號測量儀2〇擷取 信號波形曲線咖）（如_ 4所示）的數值資料，如該波形曲 線办)所對應的參數及該參數在—轉態區間上的離散數值 木σ。於本貝％例中，信號測量儀2〇爲一示波器，用於測 量主機板電壓，故® 4中座標軸的橫座標表示時間Τ，縱 座標表不電壓值V’即波形#線*)的引數參數Μ，咖爲 信號測量儀20測得的隨自變參數變化而變化的因變數。 接下來’在步驟Sl4，曲線擬合單元110利用數學方 法根據所#摘離餘值集合進行曲線擬合以求取波形 曲線吩)的最佳逼近曲線4)。所述數學方法可以爲多項式函 數法、一角函數法、對數函數法、指數函數法等等。本實 施例以多項式法爲例說明 該多項式函數的數學通式爲There is a curvature 仏 at the U ’~. If 仏> 〇, it means that point G, A curve ~ a trough in the transition interval (point (:) 2 as shown in Figure 6) Ma Ruo curve < 〇 ' indicates that the point ",) is Curve F (a peak on the ship's range (point v1 as shown in Fig. 6). Further, the positioning unit 13 turns on the peak point (n) on the curve & (e.g., point vi or point = Point, intercept 2 times the peak point (χ, ~)) the radius of curvature r (r is the reciprocal of the absolute value of the curvature F (:,)) interval interval h_R, ~ + r], within the 2R length interval The values are compared to obtain the peak position of the waveform curve in the 2R length interval and the coordinates of the position, such as the point Pm (Xm, y (Xm)), Pn (Xn, y (Xn)) in Fig. 7. 3 is a flow chart of a preferred embodiment of the non-monotonic analysis method of the signal transition interval of the present invention. First, in step s1, the signal measuring instrument 20 receives the signal from the signal generator 1 and The waveform curve (Record 13 200905208 is 〆X)) occurs in the 彳S number as 10 - the display signal changes on the display interface. In the step S12, the poor material extraction unit 1 〇〇 slave signal measurement The instrument 2 draws the signal waveform curve (as shown in _ 4), the corresponding data, and the parameter corresponding to the parameter in the - transitional interval, the value σ. In the example, the signal measuring instrument 2 is an oscilloscope for measuring the voltage of the motherboard, so the abscissa of the coordinate axis in the ® 4 indicates the time Τ, and the ordinate parameter of the ordinate is not the voltage value V', that is, the waveform #线*) The coffee is a dependent variable measured by the signal measuring instrument 20 as a function of the self-variation parameter. Next, in step S14, the curve fitting unit 110 performs a curve fitting according to the #extracted residual value set by a mathematical method. Taking the best approximation curve 4) of the waveform curve, the mathematical method may be a polynomial function method, a corner function method, a logarithm function method, an exponential function method, etc. This embodiment uses a polynomial method as an example to illustrate the polynomial function. Mathematical formula is
少W at-x = α〇 + a.Less W at-x = α〇 + a.
?-i , «爲自然數?-i , «for natural numbers
項式係數。曲線擬合單元 110求得係數矩陣W使得心與函 14 200905208 數咖)的方i D^.I):爲最小以得到最佳逼近函數 心的運算式（如圖5所示，曲線‘爲波形曲線〆χ)的最佳 逼近曲線）。 在步驟S16，計算單元12()對所求得的最佳逼近函數 U) 相對於轉態區間上的自變參數Η進行一 階與二階微分 ：^j ’亦即計算曲 線、在轉態區間上各離散點的斜率及曲率，得到最佳逼近 函數〜在轉態區間上的―階離散微分值集合及二階離 散微分值集合吆」。 隹人f步驟⑽’定位單元13G檢查所述—階離散微分值 是否有零值’若無零值則流程結束，若有零值 則表明最佳逼近函數〜在該轉態區間上具有非單調性：汽 W入步驟S2G ’定位單元130獲取擬合曲線F(上斜率 的點及該點的曲率①。若ο則表明 ’ Ά)舄曲線〜在轉態區間上的一個波谷（如圖6 15 200905208 斤丁的””占v2 ) ’右曲線化)<〇，則表明則表明點“，尸（1 ) 马曲線F⑴在轉癌區間上的一個波峰（如圖己所示的點补 在步驟S22 ’定位單元13〇以各峰值點（ά)爲 中心截取兩倍於該蜂值點曲率半徑R (R爲曲率仏絕對值 :倒„引數區間U_R，X,+R](如圖6所示)。在步驟似， 疋單7C 13〇對各卜,+ _R，x +R]區間内的數值[办^進行比 較’得到波形曲線*)在各U_R，叫區間内的峰值位置 及該位置的座標，如圖7中的點Pm(xm，綱)，Item coefficient. The curve fitting unit 110 obtains the coefficient matrix W such that the squares of the heart and the function 14 200905208) i D^.I): is the smallest to obtain the best approximation function of the heart (as shown in FIG. 5, the curve 'is The best approximation curve for the waveform curve 〆χ). In step S16, the calculating unit 12() performs first-order and second-order differentiation on the obtained optimal approximation function U) with respect to the self-varying parameter 转 on the transition interval: ^j', that is, the calculation curve, in the transition interval The slope and curvature of each discrete point are obtained to obtain the best approximation function ~ the set of discrete differential values and the set of second-order discrete differential values in the transition interval. Step (10) 'The positioning unit 13G checks whether the -order discrete differential value has a zero value'. If there is no zero value, the flow ends. If there is a zero value, the best approximation function ~ has a non-monotonicity in the transition interval. Sex: Steam into the step S2G 'The positioning unit 130 obtains the fitting curve F (the point of the upper slope and the curvature of the point 1. If ο indicates the ' Ά) 舄 curve ~ a trough on the transition interval (Figure 6 15 200905208 斤丁的"" v2) 'Right curve" &〇; 〇, it indicates that the point ", corpse (1) horse curve F (1) in the turn of the cancer zone (as shown in the point of the patch) In step S22, the positioning unit 13 截 takes twice the curvature radius R of the bee point centering on each peak point (ά) (R is the curvature 仏 absolute value: the inverted quotation interval U_R, X, +R) (eg Figure 6). In the step, the value of the 7C 13〇 for each Bu, + _R, x + R] interval [to compare ^ to get the waveform curve *) in each U_R, called the peak within the interval The position and the coordinates of the position, as shown by the point Pm (xm, outline) in Figure 7,
Pn(Xn，y(X♦在步驟S26，定位單元13〇保存原始波形曲 線灿在轉隸間的各峰值點的座财訊至㈣庫4 〇。 以上所述僅為本發明之較佳實施例而已，且已達廣泛 之使用功效，凡其他未脫離本發明所揭示之精神下所完成 ，均寻變化或修飾，均應包含在下述之中請專職圍内。 【圖式簡單說明】 —圖1是本發明信號轉態區間非單調性的分析系統較佳 貫施例的硬體架構圖。 圖2是圖1中測量分析裝置的功能模組圖。 —圖3是本發明信號轉態區間非單調性的分析方法較佳 實施例的流程圖。 圖4是從一示波器上所截取的部分信號轉態區間的原 始波形曲線圖。 八圖5中顯示的是應用本發明信號轉態區間非單調性的 刀析系、.充及方去所求取得到與圖4中的原始波形曲線的最 佳逼近曲線圖。 16 200905208 • 圖6是求取所述最佳逼近曲線的峰值點的示意圖。 ， 圖7是根據最佳逼近曲線的峰值點及該點的曲率半徑 確定原始波形曲線上峰值位置的示意圖。 【主要元件符號說明】 信號發生器 10 信號測量儀 20 測量分析裝置 30 資料庫 40 資料擷取單元 100 曲線擬合單元 110 計算單元 120 定位單元 130 17Pn (Xn, y (X♦ in step S26, the positioning unit 13 〇 saves the original waveform curve of each peak point between the transfer to the (4) library 4 〇. The above is only a preferred implementation of the present invention. In addition, and have achieved a wide range of use effects, all other changes or modifications that have been made without departing from the spirit of the present invention should be included in the following full-time. [Simplified illustration] 1 is a hardware architecture diagram of a preferred embodiment of a signal transition interval non-monotonic analysis system of the present invention. FIG. 2 is a functional block diagram of the measurement and analysis device of FIG. 1. FIG. 3 is a signal transition state of the present invention. A flow chart of a preferred embodiment of the interval non-monotonic analysis method. Figure 4 is a plot of the original waveform of a partial signal transition interval taken from an oscilloscope. Figure 8 shows the signal transition interval to which the present invention is applied. The non-monotonic knife system, the charge and the square are obtained to obtain the best approximation curve with the original waveform curve in Fig. 4. 16 200905208 • Figure 6 is the peak point of the best approximation curve Schematic diagram, Figure 7 is based on the best approximation The peak point of the line and the radius of curvature of the point determine the position of the peak position on the original waveform. [Key element symbol description] Signal generator 10 Signal meter 20 Measurement and analysis device 30 Database 40 Data acquisition unit 100 Curve fitting unit 110 computing unit 120 positioning unit 130 17