ADAPTIVE CONTROLLER DESIGN FOR CNC MACHINING PARAMETER

MAXIMIZATION
1.0

OBJECTIVE OF STUDY/PROBLEM STATEMENT

Machinability is one of the important properties of a material. It is about cutting

the material with maximum metal removal, in shortest time, with best surface finish
while having maximum tool life. The high quality of surface finish is very important in
order to face the required accuracy and marketing needs. Optimum selection of
cutting conditions importantly contribute to the increase of productivity and the
reduction of costs, therefore utmost attention is paid to this problem in this
contribution.
In conventional CNC machining, parameter such as feed rate, spindle speed
and depth of cut are calculated according to machining handbooks or the users
experience, and so the selected machining parameters are usually conservative to
avoid machining failure (such as tool wear, tool breaking, surface defect, workpiece
strength change etc). Even if the machining parameters are optimised off-line by an
optimisation
adjusted

algorithm,

during

the

they

cannot

machining

be
CAD/CAM: Design,
process planning, NC
programming

process.

Input machining
parameter manually
into CAM such as
material, feedrate,
spindle speed, depth
of cut etc.

Therefore during machining, the parameter

STEP-NC: NC code
and inspection
information as
machining models

cannot be adjust according to machining

Data transfer:
Manually Data Input
(MDI) or using DNC

condition. This could lead to tool breakage,

unintended surface roughness and machine
Setup on
machine

downtime.
With a manually controlled machine
tool, the operator watches for changes in
Machining
process

machining performance (caused, for example,

by a dull tool or a harder workpiece) and
makes

the

adjustments.

necessary
Early

work

mechanical
in

CNC

Inspection/Quality
assurance

Downtime:
Tool breakage, dull
tool, machine
breakdown etc.

machining

Figure 1 : Conventional CNC Milling Machining process flow

demonstrated that manually input data/parameters into CNC machine controllers

could not maintain system performance and stability as the force process changed.

The machining process is variable owing to tool wear, temperature change

and other disturbances. To ensure the quality of machining products, to reduce the
machining costs and increase the machining efficiency, it is necessary to optimise
the machining process at the time that the machine tools are selected for CNC
machining and during machining. The machining parameters must then be adjusted
in real-time so as to satisfy the optimal machining criteria.
The specific aims and objectives of the study are to implement :
1) Research and experiment on machining parameter involve in adaptive
machining force. Machining force control is devoted to the subject of adaptive
techniques. Data are collected and analyse according to variable parameter
such as type of tool and workpiece material (hardness), feedrate, spindle
speed, depth of cut uses of coolant, working temperature and etc.
2) The development of adaptive machining force controllers and interface within
the machine controller itself. It involves the application of various type of
sensors into the machine itself (hardware). The hardware and experiment can
be done using CNC milling traineer machine or router machine.
3) Databases of machining parameter for CNC milling are develop to integrate
with the adaptive controller/fuzzy logic.

Figure 2 : Machine are integrated with machining database using adaptive

control controller (Source : Americanmachinist.com)

2.0

LITERATURE REVIEW

Adaptive control constraint is one of the machining process control types. The major
adaptive control constraint systems are based on the feedback control, parameter
adaptive control/self-tuning control, model reference adaptive control, variable
structure control/sliding mode control, neural network control, and fuzzy control.
Their typical applications to constant cutting force control system are challenging to
establish.

Figure 3 : Process and parameter involve in CNC Milling Machining (Source :

http://www.mi.t.kyoto-u.ac.jp)

Machining force regulation is a challenging problem since the force process

varies significantly under normal operating conditions. Since fixed-gain controllers
cannot guarantee system performance and stability as the force process varies, a
substantial research effort has been invested in the development of adaptive force
controllers. However, adaptive controllers can be difficult to develop, analyze,
implement, and maintain due to their inherent complexity. Consequently, adaptive
machining force controllers have found little application in industry.
Adaptive controllers estimate force process model parameters on-line and adjust the
controller gains accordingly, thus, the need for off-line calibration tests are
eliminated. While adaptive controllers are able to account for changes in the force
process, these systems must be carefully tuned and exhibit complex, and sometimes
undesirable, behavior. Problems commonly encountered in adaptive control systems
include bursting, parameter drift, poorly scaled parameters, maintaining machining
forces at a constant level increases productivity, prevents tool breakage, regulates
tool deflections that lead to geometric workpiece errors.
4

Adaptive control system is a logical extension of the CNC-mechanism. In CNC

mechanism the cutting speed and feed rates are prescribed by the part programmer.
The determination of these operating parameters depends on the knowledge and
experience regarding the work piece, tool materials, coolant conditions and other
factors. By contrast in adaptive control machining, there is improvement in the
production rate and reduction in the machining cost as a result of calculating and
setting of optimal parameters during machining. This calculation is based on
measurements of process variables in real time and followed by a subsequent online adjustments of the parameters subjected to machining constraints in order to
optimize the performance of the overall system.

FUNCTIONS OF ADAPTIVE CONTROL

The three functions of adaptive control are:
1. Identification function.
This involves determining the current performance of the process or system.
Normally,
the performance quality of the system is defined by some relevant index of
performance. The identification function is concerned with determining the current
value of this performance measure by making use of the feedback data from the
process. Since the environment will change overtime, the performance of the system
will also change. Accordingly the identification is one that must proceed over time or
less continuously. Identification of the system may involve a number of possible
measurements activities.

2. Decision function.
Once the system performance is determined, the next function is to decide how the
control mechanism should be adjusted to improve process performance. The
decision procedure is carried out by means of a pre-programmed logic provided by
the designer. Depending upon the logic the decision may be to change one or more
of the controllable process.

3. Modification function.
The third AC function is to implement the decision. While the decision function is a
logic function, modification is concerned with a physical or mechanical change in the
5

system. It is a hardware function rather than a software function. The modification

involves changing the system parameters or variables so as to drive the process
towards a more optimal state. The process is assumed to be influenced by some
time varying environment. The adaptive system first identifies the current
performance by taking measurements of inputs and outputs. Depending on current
performance, a decision procedure is carried out to determine what changes are
needed to improve system performance. Actual changes to the system are made in
the modification function.

3.0

SCOPE OF STUDY

Adaptive control possesses attributes of both feedback control and optimal control.
Like a feedback system measurements are taken on certain process variables. Like
an optimal system,
an overall measure of performance is used. In adaptive control, this measure is
called the index of performance (IP). The feature that distinguishes adaptive control
from other two types is that an adaptive system is designed to operate in a time
varying environment. It is not usual for a system to exist in environments that change
over the course of time. An adaptive control system is designed to operate for the
changing environment by monitoring its performance and altering accordingly, some
aspects of its control mechanism to achieve optimal or near optimal routine. The
feedback and optimal systems operate in a known or deterministic environment.

Method of measuring cutting forces from the displacements of rotating spindle

shafts. A capacitance displacement sensor is integrated into the spindle and
measures static and dynamic variations of the gap between the sensor head and the
rotating spindle shaft under cutting load. To calibrate the sensing system, the tool is
loaded statically while the deflection of the tool is measured with the capacitance
probe. With this calibration, the displacement sensor can be used as an indirect
force sensor. However, the measurement bandwidth is limited by the natural modes
of the spindle structure. If cutting force frequency contents are within the range of the
natural modes of the spindle structure or higher, the measurements are distorted due
6

to the dynamic characteristics of the spindle system. In order to increase the

bandwidth of the indirect force sensor by compensating for the spindle dynamics, the
design of a Kalman filter scheme, which is based on the frequency response function
(FRF) of the displacement sensor system to the cutting force, is presented in this
paper. With the suggested sensing and signal processing method, the frequency
bandwidth of the sensor system is increased significantly, from 350 to approximately
1000 Hz. The proposed indirect force sensor system is tested experimentally by
conducting cutting tests up to 12,000 rpm with a
five-fluted end mill. Besides cutting forces, the
measured displacements can also be affected by
factors such as roundness errors, unbalance at
different speeds, or dilatation of the spindle shaft
due to temperature variations.
The adaptive controller is fed by signals of two
sensors:
I. Spindle torque sensor: The spindle torque is
measured by strain-gauges mounted on the
machine spindle.
II. Tool vibration sensor: The tool vibration is
measured by two accelerometers mounted on the
machining spindle housing.
The sensor data are fed to the adaptive
control unit which calculates new feed and speed
to optimize metal removal rate within the limits of a
set of constraints. These constraints are as follows:
1. Maximum permissible spindle speed in rpm. This limit can be imposed either by
the machine capability or from metal cutting considerations.
2. Minimum spindle speed. It takes into account limitations of the spindle drive.
3. Maximum allowed torque. This limit is prescribed by cutter radius and length as
well as by depth of cut and finish requirements.
4. Maximum allowed chip load, in /rev. the chip load is actually the feed per tooth.
The number of teeth on the cutter should fed to the control by the operator.
5. Minimum chip load. This constraint is required to prevent slow feeds.
7

6. Maximum permitted federate in rpm. This constraint takes into account the cutter
radius and length, spindle material, accuracy and finish requirement.
7. Maximum allowed vibration, measured as a percentage of the acceptable
operating range. The zero point is used to indicate air gap.
8. Impact chip load, in in/rev. This limit is the maximum cutter feed per spindle
revolution that would be allowed in traveling through unprogrammed air gaps. This
limits the feed of the tools when entering in to the work piece.

Perfomance
function

Required
force/power

Program
position
-Force
-Power
-Tool Wear
-Deflection
-Work quality
-Temperature

ADAPTIVE CONTROL LEVEL

Feed
Speed

Position

INTERPOLATION
Axial
Command

SERVO
CONTROL
LOOPS

SPINDLE
CONTROL

SIGNAL
PROCESSING

Online
Feedback

Online
Feedback

Sensed signals:
-Force
-Power
-AE
-Vibration
-Temperature
-Vision

MACHINE TOOL

Figure 5 : Online feedback and signal processing in adaptive control

4.0

METHODOLOGICAL RESEARCH

Most of existing machine learning uses example data or past experience (data
collection and database) to solve a given problem using computer programming.

Methods or technique used in machine learning incude such as statistics, pattern

recognition, neural networks, artificial intelligence, signal processing, control, and
data mining. And the analysis is veried via simulation and experimental studies.
Adaptive Algorithm for Feedrate Smoothing of High Speed Machining :
To satisfy the need of high speed machining,a look-ahead algorithm of series of
moves is proposed for the process of speed adjustment in advance, which avoids
the occurrence of overload of cutter tool on the corner. It will calculate feedrate
speed approximately, thus high resolution and fine surface quality can be obtained
while machining speed is high.The algorithm is effective and has been adopted by
CNC system of newly developed high-speed milling machine.
Neural networks (NN), based on the principles of human brain operation, have
received considerable and increasing interest over the past decade. Compared to
the traditional computing methods, an NN is robust and global. It can be used to
learn any nonlinear function and can be used for any nonlinear optimisation problem.
NN are widely used for system modelling, machine learning, function optimising,
image processing and intelligent control.
A neural network-based approach to complex optimization of cutting parameters is
proposed. It describes the multi-objective technique of optimization of cutting
conditions by means of the neural networks taking into consideration the
technological, economic and organizational limitations. To reach higher precision of
the predicted results, a neural optimization algorithm is developed and presented to
ensure simple, fast and efficient optimization of all important milling parameters.
The approach is suitable for fast determination of optimum cutting parameters during
machining, where there is not enough time for deep analysis.

Figure 6 : Process flow in adaptive machining (Source : www.witshow.org)

On-Machine Verification is another technique which uses probing equipment
on the machine tool. It allows initial checking of machined parts to be carried out in
site on the machine rather than having to transfer them to coordinate measuring
machines for inspection. The main advantage of this On-Machine Verification is that
any mistakes are discovered where they can be corrected on the machine tool.
Repeated cycles of machining and inspection, interspersed with long set-up times on
the respective pieces of equipment, are avoided, meaning that overall manufacturing
times can be reduced. On-Machine Verification can give huge time savings by
enabling the quality of the component being machined to be monitored at critical
stages in the manufacturing process. This allows any errors to be detected earlier,
and so corrected more quickly and at lower cost. For example, it will be possible to
check that the correct amount of stock has been left on the component after a
roughing operation, rather than having to wait until all machining operations have
been completed before discovering that an error has been made. Similarly, the
extent of any damage caused, for example, by a tool breakage, can be assessed
and a decision made immediately to determine whether the part can still be
completed within tolerance.
10

SOURCES OF VARIABILITY IN CNC MILLING MACHINING :

The following are the typical sources of variability to be monitor in machining where
adaptive control can be most advantageously applied.

1. Variable geometry of cut in the form of changing depth or width of cut: In these
cases,
feed rate is usually adjusted to compensate for the variability.
2. Variable work piece hardness and Variable machinability: When hard spots or
other areas of difficulty are encountered in a work piece, either speed or feed is
reduced to avoid premature failure of tool,
3. Variable work piece rigidity: if the work piece deflect as a result of insufficient
rigidity in set up, the feed rate must be reduced to maintain accuracy in the process.
4. Tool wear: It has been observed in research that as the tool begins to dull, the
cutting force increases. The adaptive controller will typically respond to tool dulling
by reducing feed rate.
These are the sources of variability present themselves as time varying and for most
part, unpredictable changes in the machining process.

5.0

WORK SCHEDULE
11

STAGE 1-Reading and Research

Proposal and field of study agreed.

1st Year
2nd Year
3rd Year
1st Semester 2nd Semester 1st Semester 2nd Semester 1st Semester 2nd Semester

Assignment of supervisor
Agreement of strict timetable for
supervisory meeting and progress
reports.
Provision of necessary research
equipment and facilities.
Agreement of proposal and thesis
work plan, research method and
timetable.
Research reading and data mining on
the field of study.
STAGE 2-Detailed Plan/Experiment
Construct a detailed plan of the
dissertation
System Modelling-Mathematical
modelling of the system based on
simulation software tools
Preparation of equipment and tools
for experiment.
Experimental and data collection.
System
Controller
DesignDevelopment of the architecture and
control methods for the adaptive
controller
Dynamic
Analysis-Simulation
and
analysis of the dynamic response of
the
system
for
the
designed
controller
Algorithm
Development
and
Controller
ProgrammingImplementation of the designed
controller for
control platform
programming
Prototype
construction-Hardware
assembly
for
a
prototype
construction
Data
analysis,
calculation
and
simulation.
Develop adaptive control on cnc
machine. Experiment and report
dissertation
2nd
experimental
and
data
collection.
Confirmation
and
assessment of 1st experiment. Data
analysis and simulation.
STAGE 3-Initial writing/First Draft
Draft the various sections of the
dissertation
Undertake additional research where
necessary
Undertake any additional editing and
research
STAGE 4-Final Draft
Check for errors
Final proof-read (by a friend or
yourself) and final editing
Get the dissertation bound
Submit your dissertation

6.0

BRIEF BIBLIOGRAPHY
12

M.P.Groover. CAD/CAM Computer aided design and Manufacturing,

Adaptive Controls Save Tools and Time | Machining / Cutting content from American
Machinist.

Americanmachinist.com/machining-cutting/adaptive-controls-save-tools-

and-time
Sungchul Jee . Adaptive fuzzy logic controller for feed drives of a CNC machine tool.
Yoram Koren.
http://www.personal.umich.edu/~ykoren/uploads/Adaptive_control_systems_for_mac
hining.pdf (1 May 2013)
Wits-Wichita Industrial Seminar 2013.
http://www.witshow.org/_userfiles/36/files/WITS%202011%20DELCAM%20SEMINA
R.pdf (1 May 2013)
R. L. Hecker, G. M. Flores, Q. Xie and R. Haran1. 2008. Servo control of machinetools: a review.
J. Balic, A new NC machine tool controller for step-by-step milling, Int. J. Adv. Manuf.
Technol. 18 (2001) 399-403.
Y. Liu, L. Zuo and C. Wang, Intelligent adaptive control in milling process,
International Journal of Computer Integrated Manufacturing, 12 (1999), 453-460.
L.A. Dobrzaski, K. Golombek, J. Kopac and M. Sokovic, Effect of depositing the
hard surface coatings on properties of the selected cemented carbides and tool
cermets, J. Mater. Process. Technol. 157-158 (2004), 304-311.
U. Zuperl, F. Cus, B. Mursec and T. Ploj, A hybrid analytical-neural network
approach to the determination of optimal cutting conditions, J. Mater. Process.
Technol. 157- 158, (2004) 82-90.
Y.S. Tarng, M.C. Chen and H.S. Liu, Detection of tool failure in end milling, J. Mater.
Process. Technol. 57 (1996) 55-61.
S J. Kopac, M. Sokovic and S. Dolinsek, Tribology of coated tools in conventional
and HSC machining, J. mater. process. Technol. 118 (2001) 377-384.
B. Sidda Reddy, G. Padmanabhan& K. Vijay Kumar Reddy, Surface roughness
prediction for CNC milling, Asian Journal of Scientific Research, no. 3,2008,pp. 256264.
YueJio, Shuting Lei, Z.J.Pei&E.S.Lee, Fuzzy adaptive networks in machining
process modelling: surface roughness prediction for milling operations, International
Journal of machine Tools & Manufacture, no. 44, June 2004, pp 1643-1651.
13

H. SharifiBidgoli, Optimization of surface roughness parameter in dry milling,

Journal of Achievements in Materials and Manufacturing Engineering, vol. 37, Dec.
2009, pp 571-557.
SittichaiKaewkuekool, P. Maramkapus&KomsonJirapattarasilp, 2009, A study of
influence factors affecting to surface roughness in stainless steel milling,
International Conference on Computer Engineering and Technology, pp. 299-302.
A.B. Abdullah, L.Y. Chia & Z. Samad, The effect of feed rate and cutting speed to
surface roughness, Asian Journal of Scientific Research, no. 1, 2008, pp 12-21.
Mohammad

IshtiyaqHossian,

Sakirtasdemir&Anayet

Patwari,

2008,

Development of an artificial neural network algorithm for predicting the surface

roughness in end mill of inconel 718 alloy, International Conference on Computer
Engineering and Technology, Kuala Lumpur, pp. 1321-1324.
L.Abhag& M. Hamedullah, Chip-tool interface temperature prediction model for
milling process, International Journal of Engineering Science and Technology, vol.
2, 2010, pp 382-392.
ASM International. Handbook Committee1990,Properties and Selection: Nonferrous
Alloys and Special-Purpose Materials, Vol 2, ASM Handbook Online, USA.
Geoffrey, C 1989, fundamentals of machining and machine tool, 2ndedition, Marcel
Dekker Inc, New York
AzlanMohdZain, HabibollahHaron&Safiansahrif, Prediction of surface roughness in
the end milling machining using artificial neural network, International Journal of
Engineering Science and Technology, no. 37, 2010, pp 1755-1768
E.O Ezugwu, W.F. Sales & J. Bonney, Modelling the correlation between cutting
and process parameters in high-speed machining of inconel 718 alloy using an
artificial network, International Journal of machine Tools & Manufacture, no. 45,
2005, pp 1375-1385.

7.0

SIGNIFICANCE OF RESEARCH
14

The main idea in adaptive control is the improvement in production rate and of
part quality by calculation and setting of the optimal variables during the machining
itself. This calculation is based upon measurements of process variables in real time
and is followed by a subsequent on-line adjustment of the machining variables
subject to constraints with the objective to optimize the performance of the overall
system.
A number of potential benefits accrue to the user of an adaptive control (AC)
machine tool. Following are some of the benefits gained from AC.
1. Increased production rates. Productivity improvement was the motivation force
behind the development of AC machining online adjustments to allow for variations
in work geometry, material and tool wear provide the machine with the capability to
achieve the highest metal removal rates that are consistent with existing cutting
conditions.
2. Increased tool life. In addition to higher production rates, AC will generally provide
a more efficient and uniform use of the cutter throughout its tool life.
3. Greater part protection. Instead of setting cutter force constraint limit on type basis
of maximum allowable cutter and spindle deflection, the force limit can be
established on the basis of work size tolerance.
4. Less operator intervention. The advent of AC machining has transferred control
over the process even further out of the hands of the machine operator and into the
hands of management via part programmer.
5. Machine on-line monitoring prevent machine component breakdown or failure

15