Ktl-icon-tai-lieu

Theo dõi mòn đối với đánh giá điều kiện cắt gọt (phần 2)

Được đăng lên bởi Anhtuan69
Số trang: 6 trang   |   Lượt xem: 382 lần   |   Lượt tải: 0 lần
Jolumnl of

ELSEVIER

Materials
Processing
Technology

Journal of Materials Processing Technology 62 (1996) 374-379

Tool WearMonitoring for OptimizingCutting Conditions
T. Obikawa a, C. Kaseda a*, T. Matsumura b, W.G. Gong a, T. Shirakashi ’
a Department
ofMehano-Aerospace
b Department
ofMechanIcal

Engineering,
Engineering,

Tokyo Institute of Technology,
2-I 2-I Ohokayama,
Meguro-ku,
Tokyo 152, Japan
Tokyo Denki University,
2-2 Kanda-nishikicho,
Chiyoda-ku,
Tokyo 101 Japan

Abstract
Tool flank wear during turning is monitored through artificial neural networks of which the input consists of the AR coefficients
representing the power spectrum of cutting force and some other parameters. The order of AR model is effectively determined by AIC.
The monitored and measured flank wear agree very well. The flank wear rate monitored is further used to adaptively revise the characteristic constants of a wear equation, by which the wear rate after the change of cutting conditions is predicted and the optimum conditions are fmally selected for a case study.
Keywords: cutting, tool wear, monitoring,

optimization

1. Introduction
In highly advanced machining systems the monitoring of
cutting states plays an important role to avoid cutting troubles
such as chatter vibration, excessivetool wear and tool breakage.
Thus it has been intensively studied as a key technology for
developing intelligent machining systems[l-5]. When a trouble
is detected by sensing acoustic emission, cutting force, acceleration, heat flux, cutting sound etc., however, what the machining
systems can do is to continue or stop cutting according to the
output information of the monitoring system. That is, such monitoring systemscan afford only one bit information (stop or continue) to machine tools and machining systems.The procedures
after the stop of machining for diagnosing and resolving the
machining troubles and for selecting better cutting conditions
still depend on experts’ experiences and heuristic knowledge.
Thus the monitoring systems which can detect the characteristic
features of gradually changing machining states should be required to avoid machining troubles by changing cutting conditions using the monitored information in advance and further to
optimize the cutting conditions under the given constraints.
In this study a monitoring system is first developed for estimating the tool flank wear quantitatively. In the system the
power spectrum of feed force and some other parameters are
...
Jolumnl of
Materials
Processing
Technology
ELSEVIER
Journal of Materials Processing Technology 62 (1996) 374-379
Tool Wear Monitoring for Optimizing Cutting Conditions
T. Obikawa a, C. Kaseda a *, T. Matsumura b, W.G. Gong a, T. Shirakashi
a Department ofMehano-Aerospace Engineering, Tokyo Institute of Technology, 2-I 2-I Ohokayama, Meguro-ku, Tokyo 152, Japan
b Department ofMechanIcal Engineering, Tokyo Denki University, 2-2 Kanda-nishikicho, Chiyoda-ku, Tokyo 101 Japan
Abstract
Tool flank wear during turning is monitored through artificial neural networks of which the input consists of the AR coefficients
representing the power spectrum of cutting force and some other parameters. The order of AR model is effectively determined by AIC.
The monitored and measured flank wear agree very well. The flank wear rate monitored is further used to adaptively revise the charac-
teristic constants of a wear equation, by which the wear rate after the change of cutting conditions is predicted and the optimum condi-
tions are fmally selected for a case study.
Keywords: cutting, tool wear, monitoring, optimization
1. Introduction
In highly advanced machining systems the monitoring of
cutting states plays an important role to avoid cutting troubles
such as chatter vibration, excessive tool wear and tool breakage.
Thus it has been intensively studied as a key technology for
developing intelligent machining systems [l-5]. When a trouble
is detected by sensing acoustic emission, cutting force, accelera-
tion, heat flux, cutting sound etc., however, what the machining
systems can do is to continue or stop cutting according to the
output information of the monitoring system. That is, such moni-
toring systems can afford only one bit information (stop or con-
tinue) to machine tools and machining systems. The procedures
after the stop of machining for diagnosing and resolving the
machining troubles and for selecting better cutting conditions
still depend on experts’ experiences and heuristic knowledge.
Thus the monitoring systems which can detect the characteristic
features of gradually changing machining states should be re-
quired to avoid machining troubles by changing cutting condi-
tions using the monitored information in advance and further to
optimize the cutting conditions under the given constraints.
In this study a monitoring system is first developed for esti-
mating the tool flank wear quantitatively. In the
system
the
power spectrum of feed force and some other parameters are
processed through artificial neural networks. Next the tool flank
wear monitoring system and an optimization system of cutting
conditions are integrated in order to utilize the monitored infor-
mation for efficient operation planning. In the integrated system,
the tool life and the optimum cutting conditions are predicted
* presently, Advance Technology Center, Yatnatake Honeywell
Co., Ltd., 134 Kobe-cho, Hodogaya-ku, Yokohama, 240 Japan
0924-0136/96/$15.00 0 1996 Elsevier Science S.A. All rights reserved
PII
0924.0136(96)02438-7
analytically using the prediction parameters of the flank wear
rate which are adaptively updated by learning the currently moni-
tored flank wear rate.
2. Monitoring of flank wear
2.1. Power spectrum of cutting force
The cutting force measured by a dynamometer is a sample
process with noise. In the steady state turning without chatter
vibration conducted using a rigid single point cutting tool, the
spectrum of the (white) noise of the dynamic cutting force
has almost the same power as that of the signal over the fre-
quency of interest as will be shown later. Thus the linear predic-
tion method (LPM) [6] on the basis of the auto-regressive (AR)
model is used to extract only the power spectrum
P, 0’0)
of the
signal by next equation:
2
(1)
where
p
is the order of the AR model,
Ep
is
2p
times the power
spectrum
P,
0) of the white noise, ak is the k-th m coefficient
and z = eJ “‘. The power spectrum
P,
(jo) is calculated using
Levinson-Durbin’s algorithm [7], which demands the minimum
amount of calculation.
This equation shows that
P, (jw)
has only
p
peaks of which
the number equals the order of AR model. Thus the AR coeffl-
Theo dõi mòn đối với đánh giá điều kiện cắt gọt (phần 2) - Trang 2
Để xem tài liệu đầy đủ. Xin vui lòng
Theo dõi mòn đối với đánh giá điều kiện cắt gọt (phần 2) - Người đăng: Anhtuan69
5 Tài liệu rất hay! Được đăng lên bởi - 1 giờ trước Đúng là cái mình đang tìm. Rất hay và bổ ích. Cảm ơn bạn!
6 Vietnamese
Theo dõi mòn đối với đánh giá điều kiện cắt gọt (phần 2) 9 10 388