基于西門子plc1200的液壓式壓力機控制系統(tǒng)設計含CAD圖

基于西門子plc1200的液壓式壓力機控制系統(tǒng)設計含CAD圖,基于,西門子,plc1200,液壓式,壓力機,控制系統(tǒng),設計,cad 外文出處： Sensors?and?Actuators?A?125(2006)?186－191 1．外文資料翻譯譯文（約3000漢字）： 用于裂紋檢測和表征的脈沖磁通泄漏技術(shù) 阿里索菲安*，桂云田，Sofiane扎伊 哈德斯大學計算和工程學院，Queensgate，Hudders實驗室HD1 3DH，英國 2004年12月21日收到; 2005年6月16日修訂; 于2005年7月15日接受 2005年8月29日在線 摘要：磁通泄漏（MFL）技術(shù)已經(jīng)廣泛用于通過施加磁化的非侵入性檢查鋼裝置。在缺陷可能發(fā)生在被檢查的結(jié)構(gòu)的近表面和遠表面的情況下，當前的MFL技術(shù)不能確定它們的大致尺寸。因此，可能必須包括額外的換能器以提供所需的額外信息。本文提出了一種稱為脈沖磁通泄漏（PMFL）的新方法，用于裂紋檢測和表征。介紹了探頭設計和方法。通過理論模擬和實驗研究了時頻域中的信號特征。結(jié)果表明，該技術(shù)可以潛在地提供關于缺陷的附加信息。最后，建議潛在的應用程序。 ?2005 Elsevier B.V.保留所有權(quán)利。 關鍵詞：脈沖磁場; NDT＆E;缺陷檢測和表征 1.介紹 磁漏技術(shù)廣泛應用于管道和油箱地板檢查[1-7]。這種技術(shù)需要被測試樣品的磁化。磁化產(chǎn)生在一定方向上在試樣中流動的磁通量，其理想地垂直于待檢測的裂紋的軸。任何缺陷的存在將實現(xiàn)為試樣中的通量的磁導率的突然變化。有缺陷的部件的磁導率通常低于無瑕疵的部件，提供高的通量阻力，并迫使它采取不同的路線。在其他路徑磁飽和的情況下，一些通量離開樣品到周圍空間，暫時導致通量“泄漏”。這種泄漏可以通過位于試樣表面附近的磁傳感器容易地檢測到。影響漏磁通分布的缺陷參數(shù)是缺陷深度與管壁厚度，長度，寬度，邊緣處的銳度和最大深度處的銳度的比率[2]。在實踐中， 磁化裝置通常是永磁體或電磁體[8,9]。對于直流檢測，霍爾器件，磁阻和SQUID [10]可用于測量漏磁場。對于交流測量，拾波線圈是另一種選擇。 MFL技術(shù)的優(yōu)點是其簡單性和低成本。與渦流技術(shù)相比，該技術(shù)對磁性材料的磁性能的變化更加魯棒，渦流技術(shù)也屬于電磁NDT技術(shù)。像許多電磁技術(shù)一樣，MFL也是非接觸的，這是在線動態(tài)檢查的非常有用的特征。然而，與渦流不同，MFL僅適用于磁性材料。 在許多應用中期望改進NDT技術(shù)的精度，例如管道檢查，其中缺陷檢測和表征的良好精度可以幫助減少不必要的昂貴的管道替換。為了滿足這一要求，本文提出了脈沖MFL技術(shù)。與其他MFL技術(shù)相比，該技術(shù)可能提供關于結(jié)構(gòu)缺陷的更豐富的信息。我們對這個提出的技術(shù)的研究提出在本文。在下面的章節(jié)中，報告了PMFL的模擬和實驗，然后得出結(jié)論和進一步的工作。 2.脈沖MFL 直流MFL技術(shù)提供關于在位置和尺寸方面檢測到的缺陷的有限信息。通常，必須確保僅存在一種類型的缺陷，并且這些缺陷只能發(fā)生在被檢查結(jié)構(gòu)的一側(cè)，以允許精確推斷缺陷尺寸，因為該技術(shù)僅依賴于一個測量特征，即磁場泄漏強度。在一些實施方式中，它們與其他模態(tài)的傳感器互補，以允許區(qū)分近表面和遠表面缺陷。對于交流MFL，檢查通常僅對樣品的一側(cè)敏感，取決于所選擇的激發(fā)頻率。使用脈沖MFL，探頭以方波驅(qū)動，豐富的頻率組件可以提供來自不同深度的信息，由于皮膚效應。預期我們可以檢查較厚的樣品的遠側(cè)缺陷，同時對近表面缺陷具有良好的靈敏度。此外，我們還應該有額外的信息，如缺陷的位置和大小。脈沖MFL系統(tǒng)的發(fā)展是我們對脈沖渦流NDT系統(tǒng)工作的延伸[11,12]。 為了探索該技術(shù)的潛力，使用U形鐵氧體磁軛設計和構(gòu)建探針。圖。如圖1所示： 圖。 （a）探針和測試充注的示意圖和（b）脈沖MFL探針布局和尺寸。 探頭的尺寸（mm）。來自霍尼韋爾的SS490系列[13]的霍爾器件傳感器已經(jīng)被選擇，并被定位在軛的極之間的中間，以測量垂直于樣品表面的磁場密度?；魻杺鞲衅鞯撵`敏度為3.125 mV / G。線圈的線圈纏繞在磁軛的頂部水平部分周圍，并以矩形波形驅(qū)動以激勵。在操作期間，控制激勵電流以避免鐵氧體磁軛磁飽和。使用100kHz采樣率的14位數(shù)字化進行數(shù)據(jù)采集。 3.模擬 有限元模型（FEM）已廣泛用于電磁無損檢測技術(shù)的研究，包括MFL [4]。在這項工作中，一個稱為FEMLAB的FEM包用于研究表面和子表面裂紋對磁場的影響，并預測系統(tǒng)輸出。該包使用有限差分法執(zhí)行我們所需的瞬態(tài)分析。圖。圖2示出了在模擬中使用的網(wǎng)格模型。圍繞插槽創(chuàng)建更精細的網(wǎng)格，以提供更準確的結(jié)果。在模擬中使用的所有槽的寬度為1mm。在本文中，表面槽的深度是指從槽的表面到底部末端的槽的長度，并且在模并且在模擬中，其從1mm變化到3mm。子表面槽的深度是樣品的頂表面與槽的頂部之間的距離，其在樣品的底表面上總是具有開口。子表面槽位于模擬中表面下0.5和1 mm處。圖。圖3示出了在每個時隙的右手側(cè)邊緣上方計算的正常磁場密度。圖。圖3a和b分別顯示了表面和子表面槽的結(jié)果。在圖中，時間= 0是激勵脈沖開始上升時。它由信號的形狀表示 圖。 3.有限元模擬結(jié)果：（a）表面槽和（b）表面下裂紋 （符號用于識別，而不是實際數(shù)據(jù)點）。 圖。 4.不同興奮的比較 技術(shù)可以潛在地通過使用信號的時間信息來區(qū)分所檢測的時隙的深度以及缺陷的位置。 圖。 圖4示出了計算的法向磁場相對于具有不同激勵波形的表面槽的到中心長軸的距離x的曲線圖。 表面槽的深度為3 mm。 曲 線顯示，瞬態(tài)的性能略好于10 kHz頻率激勵，顯著優(yōu)于直流和低頻激勵。 4.實驗結(jié)果 實驗設計給我們一些初步結(jié)果，說明該技術(shù)的能力。樣品具有深度從1至9mm變化的表面槽和具有位置深度范圍的子表面槽 1.5至7mm。槽的寬度約為1mm。 線圈由脈沖寬度為40ms的方波驅(qū)動。本節(jié)中所示的信號曲線是：初始實驗結(jié)果表明，最大信號峰值幅度是根據(jù)槽的中心長軸的不同正常距離獲得的，這取決于槽是在頂表面上還是在底表面上樣品。當槽位于表面下方時，由于場擴散，測量場更加擴展。因此，正峰到負峰距離大于槽寬度。這通過圖1中繪制的實驗結(jié)果來說明。槽的寬度約為3mm。表面槽的深度為3mm，埋入槽位于表面下1mm。樣品的厚度為10mm。通過在狹縫上以1mm步長手動掃描探針獲得結(jié)果，并且x = 0是狹縫的中心主軸。每次進行測量時，探針都不移動。對于曲線獲取信號幅度。已知使用MFL技術(shù)，磁場的極性 圖。 5.子表面和表面槽的掃描結(jié)果 圖。 6.表面槽深為1，2和3毫米的結(jié)果; 激勵脈沖的上升沿在時間= 1ms開始（符號用于識別，而不是實際數(shù)據(jù)點）。 如果傳感器在裂紋上掃描則更改。曲線顯示，使用1mm掃描步長，對于表面和子表面槽，正峰和負峰之間的距離分別為4和8mm。 從圖中可以看出。如圖5所示，信號的幅度隨著探頭相對于槽軸的相對位置而變化。從現(xiàn)在開始，當探頭定位成測量最高信號幅度時，獲得所使用的輸出信號。正峰值是假設由于對稱性，負峰值具有相同的絕對振幅。圖。圖6顯示了來自具有不同深度的表面槽的結(jié)果信號。其示出了該技術(shù)能夠通過使用信號的幅度來區(qū)分所檢查的時隙的不同深度，只要時隙的位置是已知的。應當注意，實驗信號輸出相對于時間的所有曲線都被布置成使得激勵脈沖的上升沿與時間= 1ms一致。 圖。圖7示出了對于表面和子表面槽獲得的信號的比較。它清楚地示出了子表面的信號具有不同的特性，其中它最初緩慢增加并且在某個時間點之后以更快的速率增加。換句話說，子表面槽信號的拐點發(fā)生在表面槽信號的拐點之后。通過獲取信號的一階導數(shù)可以更清楚地看到這些。 實驗上發(fā)現(xiàn)，探針不能檢測到表面下1mm的子表面槽。為了證明區(qū)分表面和子表面的能力以及能夠區(qū)分子表面不連續(xù)性的位置深度，使用具有水平長度為93mm的較大軛的另一探針。結(jié)果示于圖1。如圖8所示，這表明次表面縫隙信號的拐點再次出現(xiàn)在表面縫隙信號之后。表面信號的拐點出現(xiàn)在. 圖。 7.表面和子表面槽信號; 激勵脈沖的上升沿在時間= 1ms開始（符號用于識別，而不是實際數(shù)據(jù)點）。 大致相同的時間，而較深位置的子表面槽的拐點發(fā)生在稍后的時間，比在更靠近表面的子表面槽的情況下。 所有這些結(jié)果表明，拐點可以用于區(qū)分檢測到的時隙的深度位置。 圖。 圖9示出了信號的頻率分析。 曲線支持我們的陳述，即在頻率分析中表示為相位的時間信息對于表征缺陷是有用的。 低于50Hz的低頻分量似乎不僅區(qū)分表面和子表面之間，而且區(qū)分表面下面的子表面槽的距離。 位置鑒別似乎也可以使用200 Hz左右的頻率實現(xiàn)。 因此，很清楚，槽的距離表面的距離的確定. 面對槽位于，可以方便地使用脈沖MFL技術(shù)實現(xiàn)。 圖。 8.使用較大軛的探頭的表面和子表面槽信號; 激勵脈沖的上升沿在時間= 1ms開始（符號用于識別，而不是實際數(shù)據(jù)點）。 圖。 9.表面和子表面信號的頻率分析：（a）幅度，（b）相位（符號用于識別，而不是實際數(shù)據(jù)點）。 5。結(jié)論 已經(jīng)提出和研究了脈沖MFL技術(shù)的變體。數(shù)值分析和實驗研究表明，通過使用時頻域中的特征，包括拐點的到達時間，信號幅度和頻率分量的相位變化，PMFL顯然具有缺陷位置和尺寸的優(yōu)點。已經(jīng)表明，該技術(shù)除了給出裂紋的相對深度之外，還能夠辨別裂紋的位置。探頭設計應根據(jù)手頭的應用進行調(diào)整，因為磁軛的尺寸決定了穿透深度，通常較大的磁軛提供更深的穿透。掃描結(jié)果還顯示出在軛極之間利用磁傳感器的線性陣列的潛力，以更好地理解樣品條件。模擬結(jié)果表明，瞬態(tài)或脈沖MFL執(zhí)行表面和子表面裂紋檢查的最佳總體。 PMFL的優(yōu)點使其可以潛在地適用于鐵磁材料的許多應用，包括檢測鐵磁金屬帶中的裂紋，其中缺陷可以存在于兩側(cè)，而存取限于帶的一側(cè)。對于未來的工作，將研究PMFL用于使用特征融合技術(shù)的腐蝕檢測/表征的能力[14,15]。該模擬技術(shù)也將被改進以進一步探索該技術(shù)的潛力。 確認 作者感謝EPSRC部分資助該項目。 參考文獻 [1] B.P.C. Rao, T. Jayakumar, Discontinuity characterisation using elec- tromagnetic methods, J. Non-Destruct. Test. Eval. 2 (2) (2002) 23–29. [2] S. Mukhopadhyay, G.P. Srivastava, Characterisation of metal loss defects from magnetic flux leakage signals with discrete wavelet transform, NDT&E Int. 33 (1) (2000) 57–65. [3] W. Mao, L. Clapham, D.L. Atherton, Effects of alignment of nearby corrosion pits on MFL, NDT&E Int. 36 (2) (2003) 111–116. [4] M. Katoh, N. Masumoto, K. Nishio, T. Yamaguchi, Modeling of the yoke-magnetization in MFL-testing by finite elements, NDT&E Int. 36 (7) (2003) 479–486. [5] J.C. Drury, A. Marino, A comparison of the magnetic flux leakage and ultrasonic methods in the detection and measurement of corro- sion pitting in ferrous plate and pipe, in: Proceedings of the 15th World Conference on Non-destructive Testing, Roma, 2000. [6] C.R. Coughlin, L. Clapham, D.L. Atherton, Effects of stress on MFL responses from elongated corrosion pits in pipeline steel* 1, NDT&E Int. 33 (3) (2000) 181–188. [7] M. Afzal, S. Udpa, Advanced signal processing of magnetic flux leakage data obtained from seamless gas pipeline, NDT&E Int. 35 (7) (2002) 449–457. [8] K. Mandal, T. Cramer, D.L. Atherton, The study of a racetrack- shaped defect in ferromagnetic steel by magnetic Barkhausen noise and flux leakage measurements, J. Magnet. Magnet. Mater. 212 (1–2) (2000) 231–239. [9] M. Goktepe, Non-destructive crack detection by capturing local flux leakage field, Sens. Actuators A: Phys. 91 (1–2) (2001) 70– 72. [10] H.-J. Krause, M.v. Kreutzbruck, Recent developments in SQUID NDE, Physica C: Supercond. 368 (1–4) (2002) 70–79. [11] A. Sophian, G.Y. Tian, D. Taylor, J. Rudlin, Design of a pulsed eddy current sensor for detection of defects in aircraft lap-joints, Sens. Actuators A: Phys. 101 (1–2) (2002) 92–98. [12] A. Sophian, G.Y. Tian, D. Taylor, J. Rudlin, A feature extraction technique based on principal component analysis for pulsed eddy current NDT, NDT&E Int. 36 (1) (2003) 37–41. [13] Honeywell, SS490 Series Miniature Ratiometric Linear Solid State Sensors, 2000. [14] G.Y. Tian, A. Sophian, Reduction of lift-off effects for pulsed eddy current NDT, NDT&E Int. 38 (4) (2005) 319–324. [15] G.Y. Tian, A. Sophian, Defect classification using a new feature for pulsed eddy current sensors, NDT&E Int. 38 (1) (2005) 77–82. 傳記 Ali Sophian博士于1998年獲得BE（榮譽）學士學位，2004年獲得英國哈德斯菲爾德大學的電子工程學博士學位。他目前是哈德斯菲爾德大學計算與工程學院的研究員。他的研究興趣包括渦流NDT系統(tǒng)設計，信號處理和嵌入系統(tǒng)。自從他開始了由TWI有限公司共同發(fā)起的博士項目以來，他一直致力于電磁無損檢測系統(tǒng)。 桂云田博士于1985年和1988年分別在四川大學（中國成都）獲得計量學與儀器學士學位和精密工程碩士學位。在四川大學擔任副教授3年后，他于1995年在英國攻讀博士學位，并于1998年在德比大學獲得學位。他目前是計算機學院的讀者，工程，英國哈德斯菲爾德大學和四川大學客座教授。他在傳感器，智能儀表，非破壞性測試，數(shù)字信號處理，計算機視覺和微機電領域擁有多元化和積極的研究 系統(tǒng)（MEMS），由EPSRC，皇家學會，皇家工程學院和世界級工業(yè)資助。他在上述領域出版了100多本英文或中文的書籍和論文，并成功地監(jiān)督了幾位研究員，如博士，博士和碩士研究生。他是IEEE的高級成員，也是國際期刊和會議的定期審閱者。 So fi ane Zairi博士于1996年獲得了Monastir大學應用物理學和半導體理學學士學位（榮譽）。他于1997年在法國獲得利昂里昂INSA的微電子學碩士學位。然后，他在2001年在法國里昂的中央法學院獲得了集成傳感器的博士學位。后來，他加入了在英國格拉斯哥的斯特拉斯克萊德大學的一個文后的項目。這個由SHEFC理事會資助的項目，其中制造了涉及MOEMS和RF MEMS領域的微系統(tǒng)的多功能單元庫。 Zairi博士對嵌入式微系統(tǒng)領域和使用HDL和基于FEM的分析方法的軟件建模表現(xiàn)出極大的興趣。他目前是哈德斯菲爾德大學計算與工程學院的研究助理。  2．外文資料原文（與課題相關，至少1萬印刷符號以上）： Pulsed magnetic flux leakage techniques for crack detection and characterisation Ali Sophian ?, Gui Yun Tian, Sofiane Zairi School of Computing and Engineering, University of Hudders?eld, Queensgate, Hudders?eld HD1 3DH, UK Received 21 December 2004; received in revised form 16 June 2005; accepted 15 July 2005 Available online 29 August 2005 Abstract Magnetic flux leakage (MFL) techniques have been widely used for non-intrusively inspecting steel installations by applying magnetization. In the situations where defects may take place on the near and far surfaces of the structure under inspection, current MFL techniques are unable to determine their approximate size. Consequently, an extra transducer may have to be included to provide the extra information required. This paper presents a new approach termed as pulsed magnetic flux leakage (PMFL) for crack detection and characterisation. The probe design and method are introduced. The signal features in time–frequency domains are investigated through theoretical simulations and experiments. The results show that the technique can potentially provide additional information about the defects. Lastly, potential applications are suggested. ? 2005 Elsevier B.V. All rights reserved. Keywords: Magnetic flux leakage; Pulsed magnetic field; NDT&E; Defect detection and characterisation 1. Introduction Magnetic flux leakage techniques are widely used for pipe and tank floor inspection [1–7]. This technique requires magnetisation of the specimen under test. The magnetisation generates magnetic flux flowing in the specimen in a certain direction, which is ideally perpendicular to the axis of the crack to be detected. The presence of any flaws will imple- ment as an abrupt change of magnetic permeability to the flux in the specimen. The permeability of the flawed part is gen- erally lower than flawless parts, providing high resistance to the flux and forcing it to take a different route. In cases where the other routes are magnetically saturated, some flux leaves the specimen to the surrounding space temporarily causing flux ‘leakage’. This leakage is readily detectable by a magnetic sensor located in the proximity of the specimen surface. The defect parameters that affect the distribution of the leakage flux are the ratio of depth of the defect to the thickness of the pipe wall, length, width, sharpness at the edges and sharpness at the maximum depth [2]. In practice,the magnetisation device is usually a permanent magnet or an electromagnet [8,9]. For dc inspection, Hall devices, mag- netoresistives and SQUIDs [10] can be used to measure the leakage field. For ac measurements, pick-up coils are another alternative. The advantage of MFL techniques is its simplicity and low cost. The technique is more robust to the variation of magnetic properties in magnetic materials compared to eddy current techniques, which belong to electromagnetic NDT techniques as well. Like many electromagnetic techniques, MFL is also non-contact, which is a very useful feature for on-line dynamic inspection. Unlike eddy currents, however, MFL only works with magnetic materials. Improvement in accuracy in NDT techniques is desired in many applications, such as pipe inspection where good accuracy in defect detection and characterisation can help reduce unnecessary costly pipe replacements. To meet this requirement, pulsed MFL technique is proposed in this paper. The technique potentially offers richer information about structural defects compared to the other MFL techniques. Our study on this proposed technique is presented in this paper. In the following sections, simulation and experiments on PMFL are reported, followed by conclusions and further work. 2. Pulsed MFL The dc MFL technique provides limited information on the defects detected in terms of location and sizing. Gen- erally, it has to be ensured that only one type of defect is present and these defects can only happen on one side of the inspected structure to allow accurate inference of the defect size, because the technique only relies on one mea- surement feature, i.e. the magnetic field leakage intensity. In some implementations they are complemented with sen- sors of other modality to allow discrimination of near and far surface defects. With ac MFL, the inspection is generally sensitive to only one side of the sample depending on the excitation frequency chosen. With pulsed MFL, the probe is driven with a square waveform and the rich frequency com- ponents can provide information from different depths due to the skin effects. It is expected that we could inspect thicker samples for far side defects and at the same time has good sen- sitivity for near surface defects. In addition, we should also have additional information such as the location and size of defects. The development of the pulsed MFL system is an extension of our work on pulsed eddy current NDT systems [11,12]. To explore the potential of the technique, a probe was designed and built using a U-shaped ferrite yoke. Fig. 1 shows the dimensions of the probe in mm. A Hall device sensor from Honeywell’s SS490 family [13] has been chosen and is positioned halfway between the yoke’s poles to measure the magnetic field density normal to the sample surface. The Hall sensor has a sensitivity of 3.125 mV/G. A coil of wire is wound around the top horizontal part of the yoke and driven with a rectangular waveform for excitation. During operation, the excitation current is controlled to avoid the ferrite yoke getting magnetically saturated. Data acquisition is performed using a 14-bit digitisation at 100 kHz sampling rate. Fig. 1. (a) Illustration of the probe and a test ample and (b) pulsed MFL probe layout and dimensions. 3. Simulation Finite element modelling (FEM) has been widely used for the study of electromagnetic NDT techniques, including MFL [4]. In this work, a FEM package called FEMLAB is used to study the effects of surface and sub-surface cracks on the magnetic field and to predict the system outputs. The package uses the finite difference method to perform tran- sient analysis that is required for our purpose. Fig. 2 shows the meshed model used in the simulation. Finer meshes are created surrounding the slot to give more accurate results. The width of all the slots used in the simulation is 1 mm. In this article, the depth of the surface slots refers to the length of the slot from the surface to the bottom tip of the slot and in the simulation it is varied from 1 to 3 mm. The depth of sub-surface slots is the distance between the top surface of the sample to the top of the slots that always have an opening on the bottom surface of the sample. The sub-surface slots are located 0.5 and 1 mm below the surface in the simulation. Fig. 3 shows the calculated normal magnetic field density above the right hand side edge of the each slot. Fig. 3a and b show the results from surface and sub-surface slots respec- tively. In the figures, time = 0 is when the excitation pulse starts rising. It is shown by the shapes of the signals that the technique can potentially discriminate the depths of the slots detected and also the position of the defect by using temporal information of the signals. Fig. 4 shows the plots of the calculated normal magnetic field against the distance x to the central major axis of a sur- face slot with different excitation waveforms. The surface slot’s depth is 3 mm. The plots show that the transient per- forms slightly better than the 10 kHz frequency excitation and significantly better than both the dc and the low frequency excitations. Fig. 3. FEM simulation results: (a) surface slots and (b) sub-surface cracks (symbols are for identification, not actual data points). Fig. 4. Comparison of different excitations. 4. Experimental results The experiments are designed to give us some initial results that illustrate the capabilities of the technique. The samples have surface slots with depths varying from 1 to 9 mm and sub-surface slots with location depths ranging from 0.5 to 7 mm. The width of the slots is approximately 1 mm. The coil is driven with a square waveform with a pulse width of 40 ms. The plots of signals shown in this section are: initial experiment results show that the highest signal peak ampli- tudes are obtained at different normal distances from the slot’s central major axis depending whether the slot is on the top surface or on the bottom surface of the sample. When the slot is located below the surface, the measured field is more spread out due to field dispersion. Therefore, the positive peak to negative peak distance is larger than the slot width. This is illustrated by experimental results plotted in Fig. 5. The slots’ width is approximately 3 mm. The depth of the surface slot is 3 mm and the buried slot is located 1 mm below the surface. The thickness of the sample is 10 mm. The results were obtained by manually scanning the probe over the slots with a 1 mm step and x = 0 being the central major axis of the slots. The probe is unmoved every time a measurement is taken. The signal amplitudes are taken for the plot. It is known that with MFL techniques, the polarity of the magnetic field Fig. 5. Scanning results of sub-surface and surface slots. Fig. 6. Results with surface slots with depths of 1, 2 and 3 mm; the rising edge of the excitation pulse initiates at time=1 ms (symbols are for identification, not actual data points). changes if the sensor is scanned over a crack. The plots show that with the 1 mm scanning step used, the distances between the positive and negative peaks are 4 and 8 mm for the surface and sub-surface slots, respectively. As can be seen in Fig. 5, the amplitude of the signal varies with the relative position of the probe to the slot axis. From now on, the output signals used are obtained when the probe is such positioned that the highest signal amplitude is mea- sured. The positive peaks are taken with the assumption that the negative peaks have the same absolute amplitudes due to symmetry. Fig. 6 shows the resulting signals from sur- face slots with different depths. It shows that the technique is able to differentiate different depths of the inspected slots by using the amplitudes of the signals, provided that the location of the slot is known. It should be noted that all plots of the experimental signal output against time have been arranged so that the rising edge of the excitation pulse coincides with time = 1 ms. Fig. 7 shows the comparison of the signals obtained for both surface and sub-surface slots. It clearly shows that the signal of the sub-surface has a different characteristic where it initially increases slowly and after some point in time increases at a faster rate. In other words, the inflexion points of sub-surface slot signals happen later than those of surface slot signals. These can be more clearly seen by taking the first derivative of the signals. Experimentally it was found that the probe could not detect sub-surface slots 1 mm below the surface. To demonstrate the ability to discriminate surface and sub-surface and also to be able to discriminate the location depths of the sub-surface discontinuities, another probe with a bigger yoke having a horizontal length of 93 mm is used. The results are shown in Fig. 8, which demonstrates that again the inflexion points of the sub-surface slot signals happen later than the surface slot signals. The inflexion points for surface signals occurs at Fig. 7. Surface and sub-surface slot signals; the rising edge of the excitation pulse initiates at time=1 ms (symbols are for identification, not actual data points). approximately the same time, while the inflexion point of the deeper-located sub-surface slot happen at a later time than in the case of a sub-surface slot that is closer to the surface. All these results indicate that the inflexion point can be used to discriminate the depth location of a detected slot. Fig. 9 shows the frequency analysis of the signals. The plots support our statement that the temporal information, which is represented as phase in the frequency analysis, is useful for characterising the defects. The low frequency com- ponents, below 50 Hz, seem to discriminate not only between surface and sub-surface but also discriminate the distance of the sub-surface slots below the surface. Location discrimina- tion seems to also be achievable using the frequencies around 200 Hz. It is, therefore, clear that determ