1、二一 四 年 三 月 十五 日Measurement 40 (2007) 791796Continuous weighing on a multi-stage conveyor belt with FIR filterRyosuke Tasaki , Takanori Yamazaki, Hideo Ohnishi ,Masaaki Kobayashi, Shigeru Kurosu Department of Mechanical Engineering, Kyoto Institute of Technology, JapanDepartment of Mechanical Enginee
2、ring, Oyama National College of Technology, 771 Nakakuki, Oyama 323-0806, JapanDepartment of Engineering, Shinko Co., Ltd., 4219-71 Takasai, Shimotsuma 304-0031, JapanResearch Institute, Crotech, JapanReceived 13 January 2006; received in revised form 23 March 2006; accepted 11 May 2006Available onl
3、ine 2 June 2006AbstractToday higher speed of operation and highly accurate weighing of packages during crossing a conveyor belt has been getting more important in the food and distribution industries etc. Continuous weighing means that masses of discrete packages on a conveyor belt are automatically
4、 determined in sequence. Making the proper use of new weighing scale called a multi-stage conveyor belt scale which can be created so as to adjust the conveyor belt length to the product length, we propose a simplified and effective mass estimation algorithm under practical vibration modes. Conveyor
5、 belt scales usually have maximum capacities of less than 80 kg and 140 cm, and achieve measuring rates of 150 packages per minute and more. The output signals from the conveyor belt scales are always contaminated with noises due to vibrations of the conveyor belt and the product in motion. In this
6、paper digital filter of finite-duration impulse response (FEB) type is designed to provide adequate accuracy. The experimental results on conveyor belt scales suggest that the filtering algorithm proposed here is effective enough to practical applications. As long as spaces between successive produc
7、ts are set within a specified range, the products can be weighed correctly even if products having different lengths are transported in random manner.Keywords: Mass measurement; Continuous weighing; Conveyor belt; FIR filter* Corresponding author. Tel.: +81 285 20 2212; fax: +81 285 202884.E-mail ad
8、dresses: yamaoyama-ct.ac.jp (T. Yamazaki),oonishivibra.co.jp (H. Ohnishi), mkobayasvibra.co.jp (M.Kobayashi), kurosu_shiyahoo.co.jp (S. Kurosu).1 Tel.: +81 296 43 2001; +81 296 43 2130.2 Tel./fax: +81 296 28 4476. 0263-2241/$ - see front matter 2006 Published by Elsevier Ltd.doi:10.1016/j.measuremen
9、t.2006.05.0101. IntroductionConveyor belt scales among these are most important for the production of a great variety of prepackaged products 1. When a product is put on a conveyor belt, a measured signal from the conveyor belt scale is always contaminated by noises. Since the measured signal is usu
10、ally in the lower frequency range, a filter which will effectively cut down noises at the high-frequency end can be easily designed. If, however, the product (like a cardboard box and a parcel etc.) has a low frequency component, where the noise intensity is high, it is practically impossible to sep
11、arate the measured signal from noise. There still exist real problems for which engineering development in noise-filtering is needed. The recent techniques of dynamic mass measurement have been investigated to find a way to obtain mass of the product under dynamic conditions. The key idea of dynamic
12、 measurement is that we take into consideration the various dynamic factors that affect the measured signal in the instrument to derive an estimation algorithm 2. Ono 3 proposeda method that determines mass of dynamic measurement using dynamic quantities of the sensing element actuated by gravitatio
13、nal force. Also, Lee 4 proposed the algorithm of recursive least squares regression for the measuring system simulated as a dynamic model to obtain the mass being weighed. Successful dynamic mass measurement depends mainly on a mathematical model to achieve accurate measurement. But even the simple
14、structure of a conveyor belt scale makes it difficult to obtain the exact model. On the other hand, some filtering techniques have been applied to a signal processing for the conveyor belt scale 5. In order to reduce the influence of dynamics and to improve the accuracy of mass measurement without l
15、osing the quickness, we have proposed a simplified and effective algorithm for data processing under practical conveyor belts vibrations 69.2. Basic configuration Outline of conveyor belt scales The fundamental configuration of the conveyor belt scales may be represented schematically as shown in Fi
16、g. 1. The load receiving element is a belt conveyor supported by a loadcell at the edge of the frame. The detected signal by the loadcell is sent into a FIR digital filter through a DC amplifier. The mass of the product can be estimated as the maximum value evaluated from the smoothed signal. The si
17、mulations and the experiments are carried out under the following conditions:length of product: li = 20140 cm length of the belt conveyor: Lj (L1 = 40 cm, L2 = 40 cm, L3 = 60 cm)mass of the product: mi = 2080 kg distance between products: di = 20100 cm conveyor belt speed: v = 132m/min required accu
18、racy: 0.7% samplingfrequency: fs = 2000 Hz (sampling period: Ts =0.5 ms)The total length of the multi-stage conveyor is considered in the following patterns:L = L3(=60 cm): for the single-stage conveyor belt scale,L = L1 + L2(=80 cm): for the two-stage conveyor belt scale l,L = L2 + L3(=100 cm): for
19、 the two-stage conveyor belt scale 2,L = L1 + L2 + L3(=140 cm): for the three-stage conveyor belt scale.2.2. Minimum distance between productsThe minimum distance between products must be examined correctly by the geometrical conditions. In case of a three-stage conveyor belt scale, let the minimum
20、travelling distance which is necessary for reaching the steady state value of an output signal be S and the minimum distance which is shorter, di-1 or di, be dS. The hypothetical time changes ofa loading input can be shown in Fig. 2 under the condition that dS L - li. When the product mi is transpor
21、ted onto the conveyor belt scale, the minimum travelling distance S is necessary for measuring the mass of products accurately. As can be seen from Fig. 2, the minimum distance dS between products can be expressed by:2dS L - li + S(ds=min(di-1,di) (1)By applying actual values of the product length l
22、i and the minimum distance S (=20 cm) to Inequality (1), the minimum distance dS with respect to the product length li can be obtained diagrammatically as shown in Fig. 3. It should be noted that dS is set to be not less than 40 cm.3. Design of FIR filtersThe design of digital filters is well establ
23、ished and extensively covered in the literature. There are typically two kinds of digital filters, i.e. infinite-duration impulse response (IIR) type and finite-duration impulse response (FIR) one. After the product passes through the conveyor belt and the loading input changes to zero, the transien
24、t response in the filter should be returned to the initial state so fast 1012. For this reason, FIR filter can be considered to be adequate for conveyor belt scales.Now, we explain the procedure for designing the FIR filter. Writing the normalized frequency as = fT, where f is the frequency (Hz), an
25、d T is the sampling period (s), the desired transfer function can be expressed byThe filter Hd(ej) can be easily obtained by the well-known Remez algorithm. When the lower edge frequency (s) of the stopband width is chosen as less than 0.05 for the design of a FIR filter, it becomes generally imposs
26、ible to design since the noise attenuation effect decreases rapidly, and the sampling period T should be adjusted through the down-sampling. The data to be smoothed are extracted at every down-sampling period T(=nTm, in which n is a proper integer) from the measured data at every sampling period Tm
27、= 0.5 ms. In our case, n can be chosen as n = 4, 6 and 8 (corresponding to T = 2, 3 and 4 ms). The gain plot of the filter designed for the order M = 42 is shown in Fig. 4(a). Fig. 4(b) shows the impulse response obtained for the design specifications that p = 0.002 and s = 0.05. Also, Fig. 5(a) sho
28、ws the simulation results obtained after filtering the output signal. It can be seen that undesirable signals existing in the high-frequency range can be effectively eliminated, and by increasing T the response can be smoothed.Next, Fig. 5(b) shows an example of discrete data smoothed for the down-s
29、ampling periods is processed at every sampling period (Tm = 0.5 ms). A minor contamination of the smoothed signal by high-frequency noise becomes serious in the neighborhood of maximum point. For the sampling period T = 4 ms, the resulting frequency of the sampleddate can be determined has a periodi
30、city of f = 1/T = 250 Hz. The natural frequency of the conveyor belt is in the vicinity of 200 Hz. To reduce the effect of noise a simplest first-order low pass filter (cutoff-frequency 10 Hz) is cascaded with the FIR filter.4. Technical problems4.1. Non-uniformly distributed weightTo illustrate an
31、example of crucial problems, a combined set for three products in sequence passes over the conveyor belt scale under the condition that li = 100 cm, mi = 80 kg, and di = 60 cm. Fig. 6(a) shows the real time history of the output signals. The first, second and third signals denote output signals from
32、 conveyor belt scale L1, L2 and L3 theadded signal denotes a sum of output signals that corresponds to the mass of a product. It is clear that an exact determination of masses is impossible, because the curves have several peak values like a winding path. The simplest cause of the crooked curves is that the base plate of the product is slightly curved (or uneven) or there exists a
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1