外文翻譯--制造分析進(jìn)程數(shù)據(jù)使用快速標(biāo)記技術(shù)

1、<p>　　附錄1 翻譯原文及譯文</p><p>　　Doc No: P0193-GP-01-1</p><p>　　Doc Name: Analysis of Manufacturing</p><p>　　Process Data Using</p><p>　　QUICK TechnologyTM</p&

2、gt;<p>　　Issue：1</p><p>　　Data：20 April ，2006</p><p>　　Table of Contents</p><p>　　1Executive Summary4</p><p>　　1.1Introdution4</p><p>　　1

3、.2Techniques Employed4</p><p>　　1.3Summary of Results4</p><p>　　1.4Observations5</p><p>　　2Introdution6</p><p>　　2.1Oxford BioSignals Limited6</p><

4、p>　　3External References7</p><p>　　4Glossary7</p><p>　　5Data Description7</p><p>　　5.1Data types7</p><p>　　5.2Prior Experiment Knowledge7</p><p&

5、gt;　　5.3Test Description8</p><p>　　6Pre-processing9</p><p>　　6.1Removal of Start/Stop Transients9</p><p>　　6.2Removal of Power Supply Signal9</p><p>　　6.3Frequ

6、ency Transformation9</p><p>　　7Analysis I-Visualisation12</p><p>　　7.1Visualisation of High-Dimensional Data12</p><p>　　7.2Visualising 5-D Manufacturing Process Data錯(cuò)誤!未定義書(shū)簽。

7、</p><p>　　7.3Automatic Novelty Detection錯(cuò)誤!未定義書(shū)簽。</p><p>　　7.4Conclusion of Analysis I-Visualisation錯(cuò)誤!未定義書(shū)簽。</p><p>　　8Analysis II-Signature Analysis錯(cuò)誤!未定義書(shū)簽。</p><

8、p>　　8.1Constructing Signatures錯(cuò)誤!未定義書(shū)簽。</p><p>　　8.2Visualising Signatures錯(cuò)誤!未定義書(shū)簽。</p><p>　　8.3Conclusion of Analysis II-Signature Analysis錯(cuò)誤!未定義書(shū)簽。</p><p>　　9Analysis II

9、I-Template Analysis錯(cuò)誤!未定義書(shū)簽。</p><p>　　9.1Constructing a Template of Normality錯(cuò)誤!未定義書(shū)簽。</p><p>　　9.2Results of Novelty Detection Using Template Analysis錯(cuò)誤!未定義書(shū)簽。</p><p>　　9.3Co

10、nclusion of Analysis III-Template Analysis錯(cuò)誤!未定義書(shū)簽。</p><p>　　10Analysis IV-None-linear Prediction錯(cuò)誤!未定義書(shū)簽。</p><p>　　10.1Neural Networks for On-Line Prediction錯(cuò)誤!未定義書(shū)簽。</p><p>　

11、　10.2Results of Novelty Detection using Non-linear Prediction錯(cuò)誤!未定義書(shū)簽。</p><p>　　10.3Conclusion of Analysis IV-Non-linear Prediction錯(cuò)誤!未定義書(shū)簽。</p><p>　　11Overall Conclusion錯(cuò)誤!未定義書(shū)簽。</p>

12、;<p>　　11.1Methodology錯(cuò)誤!未定義書(shū)簽。</p><p>　　11.2Summary of Tesults錯(cuò)誤!未定義書(shū)簽。</p><p>　　11.3Future Work錯(cuò)誤!未定義書(shū)簽。</p><p>　　12Appendix A-NeuroScale Visualisations錯(cuò)誤!未定義書(shū)簽。<

13、;/p><p>　　Table of Figures </p><p>　　Test 90. From top to bottom: Ax, Ay, Az, AE, SP against time t(s)</p><p>　　Power spectra for Test 19 after removal of 50Hz power supply contributio

14、n. The top plot shows a 3-D “l(fā)andspace” plot of each spectrum. The bottom plot shows a “contour” plot of the same information, with increasing signal power shown as increasing colour from black to red </p><p&g

15、t;　　Power spectra for Test 19 after removal of all spectral components beneath power threshold</p><p>　　Az against time (in seconds) for Test 19,before removal of low-power frequency components </p>

16、<p>　　Az against time (in seconds) for Test 19, after removal of low-power frequency components</p><p>　　SP for an example test, showing three automatically-detecrmined states:S1-drilling in (shown in

17、green); S2-drill-bit break-through and removal (shown in red); S3-retraction (shown in blue)</p><p>　　Example signature of variable plotted against operating-point</p><p>　　Power spectra for te

18、st 51, frequency (Hz) on the x-axis between [0 fs/2]</p><p>　　Average significant frequency </p><p>　　Visualisation of AE signatures for all tests</p><p>　　Visualisation of Ax broad

19、band signatures for all tests</p><p>　　Visualisation of Ax average-frequency signatures for all tests</p><p>　　Novelty detection using a template signature</p><p>　　Executive Summar

20、y</p><p>　　Introduction</p><p>　　The purpose of this investigation conducted by Oxford BioSignals was to examine and determine the suitability of its techniques in analyzing data from an example

21、 manufacturing process. This report has been submitted to Rolls-Royce for the expressed of assessing Oxford BioSignals’ techniques with respect to monitoring the example process. </p><p>　　The analysis condu

22、cted by Oxford BioSignals (OBS) was limited to a fixed timescale, a fixed set of challenge data for a single process (as provided by Rolls-Royce and Aachen university of Technology), with no prior domain knowledge, nor i

23、nformation of system failure .</p><p>　　Techniques Employed</p><p>　　OBS used a number of analysis techniques given the limited timescales:</p><p>　　I-Visualisation, and Cluster Ana

24、lysis </p><p>　　This powerful method allowed the evolution of the system state (fusing all available data types) to be visualised throughout the series of tests. This showed several distinct modes of operati

25、on during the series, highlighting major events observed within the data, later correlated with actual changes to the system’s operation by domain experts.</p><p>　　Cluster analysis automatically detects whi

26、ch of these events may be considered to be “abnormal”, with respect to previously observed system behavior .</p><p>　　II-Signature represents each test as a single point on a plot, allowing changes between t

27、ests to be easily identified. Abnormal tests are shown as outlying points, with normal tests forming a cluster.</p><p>　　Modeling the normal behavior of several features selected from the provided data, this

28、 method showed that advance warning of system failure could be automatically detected using these features, as well as highlighting significant events within the life of the system.</p><p>　　III-Template Ana

29、lysis </p><p>　　This method allows instantaneous sample-by –sample novelty detection, suitable for on-line implementation.</p><p>　　Using a complementary approach to Signature Analysis, this met

30、hod also models normal system behavior. Results confirmed the observation made using previous methods.</p><p>　　IV-Neural network Predictor </p><p>　　Similarly useful for on-line analysis, this

31、method uses an automated predictor of system behaviour(a neural network predictor), in which previously identified events were confirmed, and further significant episodes were detected.</p><p>　　Summary of R

32、esults</p><p>　　Early warning of system failure was independently identified by the various analysis methods employed. </p><p>　　Several significant events during the life of the process were co

33、rrelated with actual known events later revealed by system experts.</p><p>　　Changes in sensor configurations are identified, and periods of system stability (in which tests are similar to one another) are h

34、ighlighted.</p><p>　　This report shall be used as the basis for further correlation of detected events against actual occurrences within the life of the system, to be performed by Aachen University of Techno

35、logy.</p><p>　　Observations</p><p>　　Based on this limited study, OBS are confident that their techniques are applicable to condition monitoring of the example manufacturing process as follows:&

36、lt;/p><p>　　Evidence shows that automated detection of system novelty is possible, compared to its “normal” operation.</p><p>　　Early warning of system distress may be provided, giving adequate tim

37、e to take preventative maintenance actions such that system failure may be avoided.</p><p>　　Provision “fleet-wide” analysis is possible using the techniques considered within this investigation.</p>

38、<p>　　The involvement of domain knowledge from system experts alongside OBS engineers will be crucial in developing future implementations. While this “blind” analysis showed that OBS modelling techniques are approp

39、riate for process monitoring, it is the coupling of domain knowledge with OBS modelling techniques that may provide optimal diagnostic and prognostic analysis.</p><p>　　Introduction</p><p>　　Oxf

40、ord BioSignals Limited</p><p>　　This document reports on the initial analysis conducted by Oxford BioSignals of manufacturing process challenge data provided by Rolls-Royce, in conjunction with Aachen Unive

41、rsity of Technology(AUT).</p><p>　　Oxford BioSignals Limited(OBS) is a world-class provider of Acquisition, Data Fusion, Neural Networks and other Advanced Signal Processing techniques and solutions branded

42、under the collective name QUICK Technology. This technology not only provides for health and quality assurance monitoring of the operational performance of equipment and plant.</p><p>　　QUICK Technology has

43、been extensively proven in the field of gas turbine monitoring with both on-line and off-line implementations at multiple levels: as a research tool, a test bed system, a ground support tool, an on-board monitoring syste

44、m, an off-line analysis tool and a “fleet” manager.</p><p>　　Many of the techniques employed by OBS may be described as novelty detection methods. This approach has a significant advantage over many traditio

45、nal classification techniques in that it is not necessary to provide fault data to the system during development. Instead, providing a sufficiently comprehensive model of the condition can be identified automatically. As

46、 information is discovered regarding the causes of these deviations it is then possible to move from novelty detection to diagnosis, b</p><p>　　External References</p><p>　　Accompanying document

47、ation providing further information on the data sets is available in unnumbered documents.</p><p><b>　　Glossary</b></p><p>　　AUT- Aachen University of Technology </p><

48、p>　　GMM- Gaussian Mixture Model </p><p>　　MLP- Multi-Layer Perception</p><p>　　OBS- Oxford BioSignals Ltd.</p><p>　　5 Data Description</p><p>　　The following

49、 sections give a brief overview of the data set obtained by visual inspection of the data. </p><p>　　Data types</p><p>　　The data provided were recorded over a number of tests. Each test consist

50、ed of a similar procedure, in which an automated drill unit moved towards a static metallic disk at a fixed velocity (“feed”), a hole was drilled in the disk at that same feed-rate.</p><p>　　The following da

51、ta streams were recorded during each test, each sampled at a rate of 20 KHz:</p><p>　　Ax – acceleration of the disk-mounting unit in the x-plane1 , </p><p>　　Ay- acceleration of the disk-mountin

52、g unit in the y-plane1 ,</p><p>　　Az- acceleration of the disk-mounting unit in the z-plane1 ,</p><p>　　AE-RMS acoustic emission, 50-400 KHz2,</p><p>　　SP-power delivered to the dri

53、ll spindle3.</p><p>　　Tests considered in this investigation used three drill-prices (of identical product specification) as shown in Table 1.</p><p>　　Table 1-Experiment Parameters by Test</

54、p><p>　　Note that tests 16,54,128,129 were not provided, thus a series of 190 tests are analysed in this investigation. These 190 tests are labeled as shown in Table 2.</p><p>　　Table 2 –Test indic

55、es used in this report against actual test numbers</p><p>　　Prior Experiment Knowledge</p><p>　　Normal Tests</p><p>　　AUT indicated that tests [10110] could be considered “normal pr

56、ocesses”.</p><p>　　AE Sensor Placement</p><p>　　AUT noted that the position of the acoustic emission sensor was altered prior to test 77, and was adjusted prior to subsequent tests. From inspect

57、ion of AE data, it appears that AE measurements are consistent after test 84, and so:</p><p>　　·AE is assumed to be unusable for tests [176] –the sensor records only white noise;</p><p>　　&

58、#183;AE is assumed to be usable, but possibly abnormal, for tests [7783] –the sensor position is being adjusted, resulting in extreme variation in measurements;</p><p>　　·AE is assumed to be usable for

59、tests [94190] –the sensor position is held constant during these tests.</p><p>　　Thus, the range of tests assumed to be normal [10110] should be reduced to [84110] when AE is considered.</p><p>

60、　　Test Description</p><p>　　Data recorded for during a typical test are shown in Figure 1. The duration of this test is approximately t=51 seconds. This section uses this test to illustrate a typical process

61、, as described by AUT.</p><p>　　Drill power-on and power-off events may be seen at the start and end of the test as transient spikes in SP.</p><p>　　The drill unit is then moved towards the stat

62、ic disk at the constant feed rata specified in Table 1, between t=12 and 27 seconds. This corresponds to approximately constant values of SP during that period, approximately zero AE, and very lowamplitude acceleration i

63、n x-,y-,and z- planes.</p><p>　　At t=27 seconds, the drill makes contact with the static disk and begins to drill into the metal. This corresponds to a step-change in SP to a higher lever, staying approximat

64、ely constant until t=38 seconds. During this time, AE increases significantly to a largely constant but non-zero value. The values Ax and Az increase throughout this drilling operation, while the value of Ay remains appr

65、oximately zero (as it does throughout the test).</p><p>　　At t=38 seconds, the tip of the drill-bit passes through the rear face of the disk. The value of SP increases until t=44 seconds. During this period,

66、 AE reaches correspondingly high values, while Ax and Az decrease in amplitude.</p><p>　　At t=44 seconds, the direction of the drill unit is reversed, and the drill is retracted from the metal disk. Until t=

67、46 seconds, the value of SP and AE decrease rapidly. A transient is observed in Ax and Az at t =44 seconds, with vibration amplitude decreasing until t=46 seconds.</p><p>　　At t=46 seconds, the drill-bit has

68、 been completely retracted from the metal disk, and the unit continues to be withdrawn at the feed rate until the end of the test. The value of SP decreases during this period(noting the power-off transient at the very e

69、nd of the test), while the values of all three acceleration channels and AE are approximately zero.</p><p>　　.Pre-processing</p><p>　　Removal of Start/Stop Transients</p><p>　　Assum

70、ing that normal and abnormal system behaviour will be evident from data acquired during the drilling process, prior to analysis, each test was shortened by retaining only data between the start and stop events, shown as

71、transients in SP. For example, for the test shown in Figure 1, this corresponds to retaining the period [1350] seconds. </p><p>　　Removal of Power Supply Signal</p><p>　　The 50 Hz power supply a

72、ppears with in each channel, and was removed prior to analysis by application of a band-stop filter with stop-band [4951] Hz.</p><p>　　Frequency Transformation</p><p>　　Data for each test were d

73、ivided into windows of 4096 points. A 4096-point FFT for was performed using data within each window, for Ax,Ay and Az channels. This corresponds to approximately 5 FFTs per second of data，similar to the QUICK system use

74、d in aerospace analysis， shown to provide sufficient resolution for identifying frequency-based events indicative of system abnormality.</p><p>　　For the analyses performed in this investigation, all spectra

75、l components of Ax, Ay, and Ay occurring at frequency f with power Pf below some threshold Pf<h were discarded. Time-domain signals were reconstructed by performing an inverse FFT operation on each spectral window of

76、4096 points.</p><p>　　Figure 2 shows the spectral power content of Az for Test 19 after removal of the 50 Hz power supply signal, from [021] seconds, with each FFT shown between [0 fs/2] Hz. Frequency conten

77、t throughout this test is typical for all tests: the majority of significant spectral peaks are concentrated during the drilling operation(between 14 and 21 seconds, in this test). As a hole is drilled in the metal disk,

78、 power is concentrated at higher and higher frequencies, usually reaching a highest frequency(h</p><p>　　Figuer 3 shows the same test are removal of all components with power Pf<0.1. This retains the sign

79、ificant peaks in the power spectral, whilst removing components assumed to be insignificant due to their low power. </p><p>　　Figure 4 and Figure 5 show the corresponding time-series data for Ax in test 19.

80、After removal of low-power frequency components, the time-series retains only the episodes in which significant-power vibrations were observed, which are used as the basis for detection of system abnormality by several o

81、f the analysis methods used within this investigation.</p><p>　　Figure 2-Power spectra for Test 19 after removal of 50 Hz power supply contribution. The top plot shows a 3-D “l(fā)andscape” plot of each spectrum

82、. The bottom plot shows a “contour” plot of the same information, with increasing signal power shown as increasing colour from black to red.</p><p>　　Figure 3-Power spectral for Test 19 after removal of all

83、 spectral components beneath power threshold .</p><p>　　Figure 4-Az against time(in seconds) for Test 19, before removal of low-power frequency components</p><p>　　Figure 5-Az against time(in s

84、econds) for Test 19, after removal of low-power frequency components.</p><p>　　Analysis I-Visualisation</p><p>　　This section describes the first of four analysis techniques applied to the manuf

85、acturing process data-set.</p><p>　　Visualisation of High-Dimensional Data</p><p>　　Constructing a 2-D Visualization</p><p>　　The use of large numbers of measured variables introduc

86、es problems in the visualization of the resulting data. A collection of temperatures, pressures, etc. forms a high-dimensional representation of the state of a system, but this is not readily interpreted by an operator.

87、</p><p>　　Neuroscale allows the visualization of systems that have high-dimensionality by mapping data to lower numbers of dimensions(typically two,for visual inspection). It attempts to preserve the inter-p

88、attern distances in the high-dimensional data. Data which are close together in high-dimensional space are typically kept close together in 2-D space, and data that are originally far apart remain well separated after pr

89、ojection.</p><p>　　The projection is performed using a non-linear function from the data’s k dimensional space down to 2-D for visualization purposes. In this investigation, k is 5:[Ax, Ay, Az, AE, SP] are t

90、he high-dimensional sample vectors. </p><p>　　The creation of a non-linear mapping from 5-D space to 2-D requires sample data from across the range of tests. In order to reduce the large number of available

91、sample data to a quantity suitable for constructing the mapping, a summary of the data-set is required. Each test was summarized by a number of prototype 5-D vectors using the k-means clustering algorithm(in which a lar

92、ge number of data are represented by a smaller number of prototype vectors). The non-linear mapping was trained using t</p><p>　　Automatic Test Segmentation</p><p>　　To allow the examination of

93、the 5-D data using visualization, it is convenient to divide the drilling process in to three stages, corresponding to the typical behaviour of the process described in Section 5.3.</p><p>　　A heuristic algo

94、rithm was produced to perform automatic segmentation into three episodes using the SP channel, as illustrated in Figure 6(which shows a low-pass filtered version of SP superimposed on the original signal as a red line).

95、The three states identified correspond to :</p><p>　　State S1: the approximately constant-power (or slightly decreasing) initial period of drilling;</p><p>　　State S2: the peak-power period w

96、here the drill-bit passes through the disk and is removed</p><p>　　State S3: the approximately constant-power period of retraction.</p><p>　　Note that this segmentation is only the identificat

97、ion of the times of onset and offset of each of the three described states, for the purposes of graphical display as described in the next sub-section.</p><p><b>　　公司機(jī)密</b></p><p>&l

98、t;b>　　牛津信號(hào)分析機(jī)構(gòu)</b></p><p>　　文件號(hào)：P0193-GP-01=1</p><p>　　文件名：制造分析進(jìn)程數(shù)據(jù)使用快速標(biāo)記技術(shù)</p><p><b>　　論點(diǎn)：1</b></p><p>　　日期：2006.4.20</p><p><b>

99、;　　目錄</b></p><p>　　執(zhí)行概要（文章綜述）</p><p><b>　　引言</b></p><p><b>　　引用的技術(shù)</b></p><p><b>　　結(jié)論摘要</b></p><p><b>　　觀察資

100、料、報(bào)告</b></p><p><b>　　引言</b></p><p><b>　　牛津信號(hào)分析機(jī)構(gòu)</b></p><p><b>　　引用國(guó)外的參考文獻(xiàn)</b></p><p><b>　　術(shù)語(yǔ)表</b></p><

101、p><b>　　數(shù)據(jù)描述</b></p><p><b>　　數(shù)據(jù)類(lèi)型</b></p><p><b>　　試驗(yàn)狀況簡(jiǎn)介</b></p><p><b>　　測(cè)試描述</b></p><p><b>　　預(yù)處理</b></

102、p><p>　　移除開(kāi)始、終止瞬態(tài)數(shù)據(jù)</p><p><b>　　移除電源干擾信號(hào)</b></p><p><b>　　頻率變換</b></p><p><b>　　分析處理1-可視化</b></p><p><b>　　高維數(shù)據(jù)分析</b

103、></p><p><b>　　5維機(jī)械加工數(shù)據(jù)</b></p><p><b>　　自動(dòng)信號(hào)檢測(cè)</b></p><p>　　分析方案1-可視化的結(jié)論</p><p>　　分析處理2-信號(hào)處理分析</p><p><b>　　構(gòu)建信號(hào)系統(tǒng)</b>

104、</p><p><b>　　波形分析信號(hào)</b></p><p><b>　　分析結(jié)論</b></p><p>　　分析處理3-基于模板分析的數(shù)據(jù)分析</p><p><b>　　構(gòu)建普通信號(hào)模板</b></p><p>　　使用模板分析捕獲信號(hào)的結(jié)論

105、</p><p><b>　　分析結(jié)論</b></p><p>　　分析處理5-非線性預(yù)測(cè)分析</p><p>　　基于在線預(yù)測(cè)的神經(jīng)網(wǎng)絡(luò)</p><p>　　基于非線性預(yù)測(cè)的神經(jīng)網(wǎng)絡(luò)得到的結(jié)論</p><p><b>　　非線性預(yù)測(cè)結(jié)論</b></p><

106、;p><b>　　系統(tǒng)結(jié)論</b></p><p><b>　　方法學(xué)</b></p><p><b>　　結(jié)論概述</b></p><p><b>　　前景工作</b></p><p>　　12 附錄：神經(jīng)網(wǎng)絡(luò)分析</p><p