1、PDF外文：http:/ 附录 A 附录 A 外文翻译原文 Sensorless tool failure monitoring system for drilling machines Luis Alfonso Franco-Gascaa, Gilberto Herrera-Ruiza, Roco Peniche-Veraa, Rene deJesus Romero-Troncosob, Wbaldo Leal-Tafollac Abstract It is well known that on-line tool condition monitoring
2、 has great significance in modern manufacturing processes. In order to preventpossible damages to the workpiece or the machine tool, reliable techniques are required providing an on-line response to an unexpected toolfailure. Drilling is one of the most fundamental machining operations and two of th
3、e most crucial issues related to it are tool wear andfracture. During the spindle process, the motor driver current is related to the drill condition: power consumption is higher for a worn drill incomparison to a sharp drill for the same process. This difference in power consumption can be self-cor
4、related to obtain the resultingwaveform variance to provide a merit figure for tool condition. This paper describes a driver current signal analysis to estimate the toolcondition by using the discrete Wavelet Transform in order to extract the information from the original cutting force, and through
5、anautocorrelation algorithm evaluate the tool wear in the form of an asymmetry weighting function. The current is monitored from the motordriver to give a sensorless approach. Experimental results are presented to show the algorithm performance, a complete sensorless tool failuresystem which allows
6、the detection of tool failure as a function of spindle current in real time. Keywords: Tool failure； Wavelet transform； Tool monitoring 1. Introduction Cutting tools represent the highest cost in the productionprocess of the manufacturing 东北大学毕业设计（论文） 附录 A industry besides rawmaterials.
7、Drilling is one of the most common operationsin machining, thus the number of drilling machines in useperform a considerable amount of work every year 1. As inany other cuttingprocesses, tool fracture and wear arepresent in several forms； therefore, they could damage theprocessand increase the produ
8、ction costs. To reduceexpenses on the workpiece and machinery, on-line toolcondition monitoring is mandatory 2. Around 80% of thereported cases return the investment inmonitoring systemsin one month or less 3. Jantunen 4 describes the studied and applied techniquesof indirect monitoring in drilling.
9、 Vibration and sound basedworks have been reported. Such methods are very sensitiveto surrounding noise, which is inherent to the cuttingprocesses, as the acoustic and ultrasonic vibrations. Thetechniques more frequently reported are related to torque,force and feed rate, since cutting forces increa
10、se when thetool wear increases 4,5.Spindle motor current and feed driver current are closelyrelated to the forces involved in machining similar to torquemeasurement 4, since both show the amount of consumedpower in the cutting process. Driver current monitoring isthe best approach to acquire signals
11、 without sensors becausein this way the machine is not modified, even when currentsensors are used 6. In addition, the current fromservomotors is available in almost all modern rotatingmachinery as turning, milling and drilling, directly from theservodriver. Practically all the current analysis base
12、dalgorithms reported use a sensor to obtain the cuttingsignals. 2. Theoretical background According to Altintas 7, the main components ofcutting forces (dFt, dFf, dFr) can be evaluated in the x, y, andz directions. The total thrust and drill torque can beevaluated by the sum of the contribution of a
13、ll M lipelements. The total thrust force is found by adding thecontributions of the chisel and lips that can be seen as asinusoidal function plus a constant.Direct measurements of spindle current signal showsevere interference from different sources represented in theadditive model (Fig. 1). The app
14、roach of Romero-Troncoso8 is a general procedure to estimate the main componentsof the cutting force which extracts the filter characteristics.By applying this approach to the drilling problem we use alow-pass filter (LPF) to ensure that the spectral contents ofthe cutting 东北大学毕业设计（论文） 附录 A signal a
15、re preserved, while spurious data areminimized. The designed filter does not eliminate allspurious components, but a subsequent Discrete WaveletTransform (DWT) will enhance the cutting force signal byits filter bank property.The wavelet transform brings a time-frequency representationof a signal in
16、decimated form depending on theapplication detail level； the result will be given as time domainsamples at the decimated frequency in a compressedform 12. The bases of the Wavelet Transform are thewavelets, generated from a basic wavelet function bydilations and translations. Given a time-varying si
17、gnal f(t), awavelet transformation consists of computing coefficientsthat are the inner products of the signal and a family ofwavelets 9. By DWT we understand the continuouswavelets with the discrete scale and translation factors 10. The DWT is defined as stated in Eq. (1), where cj,k iscalled the w
18、avelet coefficient. This may be considered as atime-frequency map from the original signal f(t). Anapproach of multi-resolution analysis is used on the discretescale function, defined together with Eq. (3). dttfc kjkj , )(
19、 (1) )22(2 2, j jjkj ktc (2) dtttfd kjkj )()( , (3) where dj,k is called the scale coefficient and
20、 is the sampledversion of the original signal. The DWT computes thewavelet coefficients cj,k and dj,k (jZ1,., J) given by Eqs. Fig A.1 Additive model for driver current contibutors (4) and (5). n jjkj knhnxc 2, (4) And njjkj kngnxd 2, (5)
