Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract
The paper presents an intelligent system for on-line monitoring and optimization of the cutting process on the model of the ball-end milling.
An intelligent system for monitoring and optimization in ball-end milling is developed both in hardware and software. It is based on a PC,
which is connected to the CNC main processor module through a serial-port so that control and communication can be realised. The monitoring
system is based on LabVIEW software, the data acquisition system and the measuring devices (sensors) for the cutting force measuring. The
system collects the variables of the cutting process by means of sensors. The measured values are delivered to the computer program through
the data acquisition system for data processing and analysis. The optimization technique is based on genetic algorithms for the determination
of the cutting conditions in machining operations. In metal cutting processes, cutting conditions have an influence on reducing the production
cost and time and deciding the quality of a final product. Experimental results show that the proposed genetic algorithm-based procedure for
solving the optimization problem is effective and efficient, and can be integrated into a real-time intelligent manufacturing system for solving
complex machining optimization problems.
2005 Elsevier B.V. All rights reserved.
0924-0136/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2005.04.041
F. Cus et al. / Journal of Materials Processing Technology 175 (2006) 9097 91
Fig. 1. Intelligent system for on-line monitoring and optimization in ball-end milling.
convergence speed. The operating domain is defined and When the tool is cutting the workpiece, the force will
changed to be around the optimal point in its evolutionary be applied to the dynamometer through the tool. The
processes so that the convergence speed and accuracy are piezoelectric quartz in the dynamometer will be strained
improved. and an electric charge will be generated. The electric
charge is then transmitted to the multi-channel charge
amplifier through the connecting cable. The charge is then
2. On-line monitoring system amplified using the multi-channel charge amplifier. In
the multi-channel charge amplifier, different parameters
This paper presents our research progress on practical can be adjusted so that the required resolution can be
methods for the optimization of cutting conditions during achieved.
CNC machining. The on-line monitoring system for the cut- Essentially, at the output of the amplifier, the voltage will
ting force monitoring presents the data acquisition system, correspond to the force depending on the parameters set in the
LabVIEW software, and the results measured cutting forces. charge amplifier. The interface hardware module consists of a
The data acquisition system used in this experimental model connecting plan block, analogue signal conditioning modules
consists of dynamometer, fixture module, hardware and and a 16 channel A/D interface board (PC-MIO-16E-4). In
software module as shown in Fig. 2. the A/D board, the analogue signal will be transformed into
A significant amount of research has been based around a digital signal so that the LabVIEW software is able to read
the monitoring of cutting forces [38]. Force monitoring and receive the data.
is commonly taken using a table-mounted dynamometer The voltages will then be converted into forces in X, Y and
during machining. These dynamometers measure the cutting Z directions using the LabVIEW program. The LabVIEW
force in three mutually perpendicular directions notationally data acquisition module is based on a PC computer, and is
the X, Y and Z axis. The dynamometer is clamped between a general-purpose programming system with an extensive
the workpiece and the table or pallet. library of functions and subroutines for any programming
The dynamometer system is composed of a dynamometer task. It also contains an application specific library for data
(Kister Model 9259A), a multi-channel charge ampli- acquisition, serial instrument control, data processing, anal-
fier (Kister Model 5001) and their connecting cable. ysis presentation and storage.
92 F. Cus et al. / Journal of Materials Processing Technology 175 (2006) 9097
z = Rb (1)
tan b
and the axial position Z, and b is the helix angle of the cutting Z axis from the center of the hemispherical part to the point
edge of the milling cutter. on the cutting edge.
For the milling cutters of constant length the local helix The generalized equation for the chip thickness is as fol-
angle changes with respect to the milling cutter radius and it lows:
is calculated according to the equation [9]:
hb (i, j, k) = fzb sin[B(i, j, k)] sin[(i)] (10)
R()
tan b () = tan b (2)
Rb where dz is the thickness of axial differential elements.
Geometry of the ball-end milling cutter and orientation of
where R() is the tool radius in XY plane with respect to
the cutting edge are used in the equation for determination of
angle
cutting forces.
R() The equation for the tangential cutting force, radial cutting
= arcsin (3)
Rb force and axial cutting force is:
where is the angular position in the direction of Z axis from dFT,R,A = KT,R,A hb db = KT,R,A fzb sin B sin db (11)
the center of the hemispherical part to the point on the cutting
edge. where KT is the tangential coefficient of material, KR the
The radius of the cutting edge in the XY plane, which radial coefficient of material, KA the axial coefficient of mate-
touches the point on the helical and spherical cutting edge rial, dz the differential length of axial differential elements
with angle is determined as follows: and db is the differential length of cutting edge.
If instead of db, we enter [9]:
R() = 1 ( cot b 1)2 Rb (4)
dz
Angular spacing between the cutting edge on the milling db = (12)
sin
cutter [9]:
we obtain:
360
P = (5)
Nf dFT,R,A = KT,R,A fzb sin B dz (13)
where Nf is the number of cutting edges. The generalized equation for the tangential, radial and
Angular position of cutting edge: axial cutting force is:
P
(j) = j , j = 1, 2, . . . , N (6) dFT,R,A (i, j, k) = KT,R,A fzb sin[B(i, j, k)]dz (14)
N
where N is the number of angular positions, (j) the angu- The forces expressed in the Cartesian coordinate system
lar position of cutting edges and P is the angular spacing are obtained if the transformation matrix [T] is inserted [9]:
between cutting edges.
Thickness of axial differential elements on the cutting {dFX,Y,Z } = [T ]{dFR,T,A } (15)
edge of the milling cutter:
sin sin B cos B cos sin B
AD [T ] = sin cos B sin B cos cos B (16)
dz(i) = i , i = 1, 2, . . . , Nz (7)
Nz cos 0 sin
where AD is the axial depth, RD the radial depth and Nz is the
Nf
number of axial differential elements on the cutting edge of
the milling cutter. [dFX,Y,Z (i, j)] = [T ][KR,T,A ]fzb sin[B] dz (17)
Angular position of the cutting edge during cutting B(i, j, k=1
k): The total force on the cutting edge in case of jth position:
z
B(i, j, k) = (j) + P (k 1) tan b (8) Nf
Nz
Rb
Genetic algorithms are a family of computational models This operator, second among the genetic operators, is
inspired by evolution. These algorithms encode a potential mostly responsible for the progress of the search. It swaps
solution to a specific problem on a simple chromosome- the parent strings partially, causing offspring to be gener-
like data structure and apply recombination operators to ated. In this, a crossover site along the length of the string is
these structures so as to preserve critical information. In selected randomly, and the portions of the strings beyond the
a broader usage of the term, a genetic algorithm is any crossover site are swapped.
population-based model that uses selection and recombina-
tion operators to generate new sample points in a search 4.3. Mutation
space. Researchers largely working from an experimental
perspective have introduced many genetic algorithm mod- It is one of last GA operators, this is the occasional random
els. Many of these researchers are application oriented and alteration (with a small probability) of the value of a string
are typically interested in genetic algorithms as optimization position. In binary strings, this simply means changing 1 to
tools. Modelling, machining, selection of cutting conditions 0 or vice versa.
and monitoring often have to deal with the problem of opti-
mization.
A genetic algorithm was applied to the simulation model 5. Optimization of cutting conditions with GA
to determine the process parameter values that would result in
the simulated cutting forces in ball-end milling. Most of the Process modelling and optimization are two important
researchers have used traditional simulation techniques for issues in manufacturing. The manufacturing processes are
solving machining problems [10]. The traditional methods of characterized by a multiplicity of dynamically interacting
simulation and search do not fare well over a broad spectrum process variables. Cutting forces have been important fac-
of problem domains. Traditional techniques are not efficient tors to predict machining performances of any machining
when practical search space is too large. These algorithms are operation. The predictive modelling of machining operations
not robust. Numerous constraints and number of passes make requires detailed prediction of the boundary conditions for
the machining simulation problem more complicated. Tradi- stable machining.
tional techniques such as geometric programming, dynamic In a traditional CNC system, machining parameters are
programming, branch and bound techniques and quadratic usually selected at the start according to handbooks or peo-
programming found it hard to solve these problems. And ples experiences, and the selected machining parameters are
they are inclined to obtain a local optimal solution. GA comes usually conservative so as to avoid machining failure. Even if
under the class of the non-traditional search and simulation the machining parameters are optimized off-line by an opti-
techniques. mization algorithm, they cannot be adjusted in the machining
Genetic algorithms are search algorithms for simulation process, but the machining process is variable owing to tool
and optimization, based on the mechanics of natural selection wear, heat change and other disturbances. To ensure the qual-
and genetics [10]. The power of these algorithms is derived ity of the machined products, to reduce the machining costs
from a very simple heuristic assumption that the best solu- and to increase the machining efficiency, it is necessary to
tion will be found in the regions of solution space containing optimize and control the machining process on-line when the
high proposition of good solution, and that these regions can machine tools, are used for CNC machining. The machining
be identified by judicious and robust sampling of the solution parameters must be adjusted in real-time so as to satisfy some
space. The mechanics of genetic algorithms is simple, involv- optimal machining criteria.
ing copying of binary strings and the swapping of the binary Intelligent manufacturing achieves substantial savings in
strings. The simplicity of operation and computational effi- terms of money and time if it integrates an efficient automated
ciency are the two main attractions of the genetic algorithm process-planning module with other automated systems such
approach. The computations are carried out in three stages as production, transportation, assembly, etc. Process plan-
to get a result in one generation or iteration. The three stages ning involves determination of appropriate machines, tools
are reproduction, crossover and mutation. for machining parts, cutting fluid to reduce the average tem-
perature within the cutting zone and machining parameters
4.1. Reproduction under certain cutting conditions for each operation of a given
machined part. The machining economics problem consists
This is the first of the genetic operators. It is a process in in determining the process parameter, usually cutting speed,
which copies of the strings are copied into a separate string feed rate and depth of cut, in order to optimize an objec-
called the mating pool, in proportion to their fitness values. tive function. A number of objective functions by which
This implies that strings with higher fitness values will have a to measure the optimality of machining conditions include:
higher probability of contributing more strings as the search minimum unit production cost, maximum production rate,
progresses. maximum profit rate and weighted combination of several
F. Cus et al. / Journal of Materials Processing Technology 175 (2006) 9097 95
objective functions. Several cutting constraints that should were got from the analytical model and estimation, were used
be considered in machining economics include: tool-life con- for the optimal objective function and conditions as the fol-
straint, cutting force constraint, power, stable cutting region lowing.
constraint, chiptool interface temperature constraint, sur- Objective function:
face finish constraint, and roughing and finishing parameter 1 model
model
relations. min(E) = FXmax FXallowed
max
+ FYmax FYallowed
max
3
The main objective of the present paper is to determine the
optimal machining parameters that minimize the unit produc- + FZmodel
max
FZallowed
max
(20)
tion cost without violating any imposed cutting constraints.
or
5.1. Optimal working condition selection model allowed
min(E) = Fmax Fmax (21)
tools, workpieces and different machining tasks. Fmax are the estimated maximum cutting forces calculated
model
In the paper, the tool cutting conditions were optimized by the analytical cutting force model.
based on the minimum machining time that is to find the max- The tool working condition optimization program was
imum feed rate that is able to meet the tool life requirement made in Matlab. In the program, the tool cutting conditions
in the specific machining task. It is known how many cut- can be selected automatically according to the required tool.
ting millimeters are required in the specific machining task.
Approximated tool life can be estimated depended on how 5.2. An illustrative example
many tools will be used in the task. Referred to tool life esti-
mation the maximum feed rate can be determined with the Many experiments were conducted to evaluate the valid-
other possible working conditions (for example, maximum ity of the optimization model in various cutting modes and
spindle speed of the machine tool) by the analytical cutting conditions (Fig. 5). The instantaneous cutting force signals
force model. in three orthogonal directions were measured by a table-
During the machining process, we collect cutting force mounted piezoelectric dynamometer (Kistler Model 9259A).
data, which can then be compared to model predicted force These signals were amplified (Kistler Model 5001), digi-
[11]. Model parameters can then be refined to more accu- tized (PC-MIO-16E-4) and stored in computer. The measured
rately optimize the cutting conditions. In this way, the mod- data was processed with the computer program made by
els can be calibrated on-line for the specific workpiece LabVIEW. The experiments were run on the NC milling
materialcutting tool combination. machine (MORI SEIKI FRONTIER-M) and performed on
The experimental data was utilized to build the analyt- material Ck 45 and Ck 45 (XM) with improved machining
ical and genetic force model and empirical tool wear by properties. The solid ball-end milling cutter type R216.44-
genetic programming method. This analytical force model 10030-040-AL10GGC 1010 with four cutting edges, of
was taken as objective function and was optimized using a 10 mm diameter and 30 helix angle was used for machining
genetic algorithmic approach to obtain the optimal machining of the material.
parameters for the required task. A new local search opti- For the determination of optimal cutting conditions the
mization based on genetic algorithm approach is developed optimization of two variables (feeding fz and cutting speed
to solve the machining optimization model (Fig. 4). Vc ) was used. The evolutionary parameters for the genetic
The PC, which is connected to the CNC main processor algorithm were: population size 500, number of generations
through a serial-port, sends the optimal cutting conditions to 30 and number of genes of each chromosome 10. The genetic
the tool machine. The tool wear, breakage or chatter is recog- operations crossover and mutation were used. Probability of
nised by a PC supervisory control software that monitors the crossover was pc = 0.65 and mutation pc = 0.1.
process. Therefore, one of the critical phases of production Optimal cutting conditions were found in 13 generation
automation is to monitor the cutting tool on-line and ensure with average error 0.28% (Fig. 6).
timely replacement of the cutting tool if necessary. With the optimal cutting conditions the machining time
The genetic algorithm was used to optimize the tool cut- was reduced for 16.4%.
ting conditions with the analytical cutting force model. The The present model provides excellent optimization of the
maximum cutting forces of the cutting force profiles, which cutting conditions. It accurately predicts fine details of the
96 F. Cus et al. / Journal of Materials Processing Technology 175 (2006) 9097
7. Conclusion