Adaptive control-systems-paper-presentation-100115092600-phpapp01 (1)

www.studentyogi.com www.studentyogi.com

co om
Adaptive control system with knowledge server system

m
in CNC system

gi. .c
oogi
ntyy
eent
ADAPTIVE CONTROL SYSTEM WITH KNOWLEDGE
t t dd

SERVER IN INTELLEGENT CNC SYSTEM
ssuu
w. .
w

ABSTRACT:
ww
ww

In an ideal scenario of intelligent machine tools the human mechanist
was almost replaced by the controller. During the last decade many efforts have been
made to get closer to this ideal scenario, but the way of information processing within the
CNC did not change too much. The paper summarizes the requirements of an intelligent
CNC evaluating the advancement of technology in this field using different adaptive
control systems. In this paper a low cost concept for artificial intelligence named


Knowledge Server for Controllers (KSC) is introduced. It allows more devices to solve
their intelligent processing needs using the same server that is capable to process
intelligent data. The KSC concept is used in an open CNC environment to build up an
intelligent CNC.

om
Key words: Intelligent CNC, knowledge server, adaptive control systems

i.c
og
nty
de

1. INTRODUCTION:
stu

There are many definitions of the intelligent machine tools. In a
well known book Wright and Bourne said that “We must therefore acknowledge that the
degree of intelligence can be gauged by the complexity of the input and/or the difficulty
w.

of ad hoc in-process problems that get solved during a successful operation. Our
unattached, fully matured intelligent machine tool will be able to manufacture accurate
aerospace components and get a good part right the first time”. They told that an
ww

intelligent machine tool had the CAD data, the materials and the set-up plans as inputs
and could produce correctly machined parts with quality control data as outputs.
2
It is clear that adaptive control techniques are necessary to apply if one wants
intelligent CNC machine, but - of course - the usage of them is not adequate in intelligent
behavior.



Table 1 summaries the features of an intelligent CNC (Wright and Bourne
collected them more than ten years ago) and shows two further things: the positive
changes done in the recent years and the still existing gaps where - according to the
scientific community – adaptive control systems offers solutions with its information

om
processing methods.

Analyzing the above list, it is clear that many features do not require direct
adaptive control methods. We can state that the main reasons of the advancement were:

i.c
(1) The development of the hardware elements (more sensitive sensors, more precise
actuators, quicker and stronger computers etc.) even in higher requirements.

og
(2) The development of the software and the methodology mainly in the preparation
phases of the manufacturing (in design, planning, scheduling, resource management etc.)
and in the user interface issues (more comfortable and informative 'windows-like' screens
nty
and Menus).
de

Table 1. Commercial needs for the intelligent machine tools:
stu

Features (forecasted in 1988) Big Artificial
advance intelligence
by 2001 methods
still needed
w.

1. Reduce the number of scrap
parts following initial setup.
ww

2. Increase the accuracy with
which parts are made.
3. Increase the predictability of
machine tool operations.
4. Reduce the manned operations
in the machine tool
environment.


5. Reduce the skill level required
for machine setup and
operations.
6. Reduce total costs for part
fabrication.

om
7. Reduce machine downtime.
8. Increase machine throughput.
9. Increase the range of materials
that can be both setup and

i.c
machined.
10. Increase the range of possible

og
geometries for the part
11. Reduce tooling through better
operation planning
nty
12. Reduce number of operations
required for setup
13. Reduce setup time bydesigning
parts for ease of setup
de

14. Reduce the time between part
design and fabrication
stu

15. Increase the quantity of
information between the
machine
16. control and part design
w.

operations
ww

Even there is a big advance in the technology of the CNCs, the
Knowledge processing and other adaptive control methods have not appeared within the
Intelligent Open CNC System Based on the Knowledge Server Concept 3controllers, so
in some points there is no real development. Special heuristic rules, problem-solving
strategies, learning capabilities and knowledge communication features are still missing


from the recent controllers available on the market. It is also true for many new, open or
PC-based CNC s, where DSP add-on-boards provide the necessary computation power
and speed. Further requirements of intelligent CNC s can be find out and defined.

(Table 2.): The second column indicates whether the different adaptive control

om
techniques (mainly rule base systems, neural nets and fuzzy logic) would provide
methods and solutions. One can find positive answers to all these issues in the resent
literature:

i.c
og
nty
de

Table2. Future requirements of an intelligent CNC:
stu

Table 2. Future requirements of an intelligent CNC
w.

Further features Artificial
intelligence
methods
ww

would help


• Model based on-line path generation
• Automatic tool selection
• Technological based settings of the operational parameters
• Automatic compensation of machine limits
• Automatic back-step strategies

om
• Detection and compensation of geometrical deflection
• On-line selection of control algorithms
• Intelligent co-operation with other devices to solve problems

i.c
together
• Detection and correction of tool wear and breakage
• Automatic handling of rejected work pieces

og
• Detection and managing of emerging situations of the machine
tool
• Complex self-diagnostics
nty
The list may be continued with the learning capabilities and others. In the
users' point of view these features rough in a controller, that "recognizes the problem" and
de

"efficiently and reasonable solves them" with minimal disturbance of the environment of
the controller.
stu
w.

2. RESULTS IN INTELLIGENT CNC s:
ww

On the one hand in the recent literature one can find many different topics
related to intelligent CNCs. Unfortunately they often do not mean intelligent behaviors
but the application of intelligent methods. Sometimes it is the case that authors call their
devices “intelligent" if one module of the system contains based method. On the other
hand (2) the key controller vendors leave everything to the users or machine tool builders
offering PC/Windows based CNCs. With these systems any software modules (e.g. even


adaptive control based ones) can be coupled into the controller but they do not offer J real
solutions or methodologies, but only software possibilities.

The following list summarizes the most important active research topics in this
field. A real intelligent CNC would contain most of these issues.

om

1) Fuzzy logic based concurrent control of some operating parameters
(E.g. cutting speed, depth of cut, feed rate) independently from the
Given tool and the work piece.

i.c
2) Neural nets and fuzzy rules in the CNC's control algorithms.
3) Optimal path planning, real-time correction of the trajectory.

og
4). Compensation of temperature (and other) deformations.
5) Life time management of the tools and other parts of the machine
Tool including self-diagnostics.
nty
6) Tool breakage detection (maybe forecasting) and tool wear
Monitoring (maybe compensation) with AI methods.
7) The utilization of CNC management (setup, orders, etc.) via
Intelligent agents.
de

Intelligent parts of a CNC can be classified into three groups, namely:
stu

(1) Tool monitoring,
(2) operation/machine tool modeling and
(3) Adaptive control.
w.

A general problem in all the three groups is that the adaptive control based
solutions are typically limited and valid only in a very narrow field. If one changes some
parameters of the operation or the environment, the earlier successful methods become
ww

false.
A special type of adaptivity partly helps on this hard and well-known problem.
If it is possible to replace the different modules of the controller time by time, than one
can guarantee, that a given adaptive control module can run within its limitation, and over
it another module (e.g. a much simpler one) covers the same functionality. It can be
realized (among others) if the controller is open to allow this replacement.


3. Adaptive Control System :
In adaptive control, the operating parameters automatically adapt themselves to
conform to new circumstances such as changes in the dynamics of the particular process
and any arise.
The adaptive would check load conditions, adapt an appropriate desired braking

om
profile (for ex: antilock brake system and traction control), and then use feed back to
implement it. With advanced adaptive controllers the gain may vary continuously with
changes in operating condition.

i.c
Purpose of adaptive control:
1. to optimize production rate
2. to optimize product cost

og
3. to minimize cost
The functions common to adaptive control systems are the following:
1. Determine the operating conditions of the process, including measures of
nty
performance. Thos typically achieved by using sensors which measure process
parameters (such as force, torque, vibration, and temperature).
2. Configure the process control in response to the operating conditions. Large changes
de

in the operating conditions may provoke a decision to make a major switch in control
strategy. More modest alterations may be the modifications of process parameters
(such as changing the speed of operation or the feed in manufacturing).
stu

3. continue to monitor the process, making further changes in controller when and
As needed.
In an operation such as turning on lathe, the adaptive control system senses real –
w.

time cutting forces, torque, temperature, tool-ware, tool chipping or tool fracture, and
surface finish of the work piece. The system then converts this information into
commands that modify the process parameters on the machine tool to hold them
ww

constant (or with in certain limits) or to optimize the cutting operation.

4. DIFFERENT APPROACHES OF KNOWLEDGE
SERVERS:


The features of World Wide Web led to introduce knowledge server to easier
solve the installation and version control problems of expert systems and to provide a web
based interface of the knowledge base for the different users. Some advanced knowledge
based systems are based on this concept. There are some applications of knowledge
servers in manufacturing. In the HPKB (High Performance Knowledge Environment)

om
some hundred thousand rules are performed in an intelligent knowledge environment. In
this project the different intelligent components are called knowledge servers. The
components are communicating with each other via the OKBC (Open Knowledge Base
Connectivity) protocol specified at Stanford.

i.c
On communication networks, the protocol named MAP uses a bus configuration,
broad band transmission, a token passing access scheme and data Transmission rate of

og
10mbps. MAP is based on specification defined by the identical standard organizations

(ISO) called the open system interconnection (OSI) reference model.
nty
The seven layer structure of MAP standard the first four layers are connected with
the inter connection functions, and the top three layers are connected with inter working
functions. The application layer (seventh layer) is the highest layer in MAP at the time of
de

this writing. It is possible that this layer may be sub divided into multiple layers as
applications of communication protocol and computer techniques evolve in future.
stu
w.
ww

5. KNOWLEDGE SERVER FOR CONTROLLERS:

Knowledge Server for Controllers (KSC) is defined as a server providing
capability of intelligent data processing for other systems. It allows the basic system to
reach external intelligent processing resources, because it does not have any. The KSC
contains a high performance reasoning tool, and different knowledge based modules. All


the modules have their special rules and procedures. The client system calls these
modules, passes them specific data if necessary, and the KSC module can collect data if
the knowledge processing requires. All the data acquisition and user interaction is done
by the client system. It is clear that in KSC the clients have much more tasks than a
simple browser based user interface and in the applications listed in the previous chapter.

om
It should be stated that KSC does not deal with fuzzy and neural net based
adaptive control modules. The computing power and the necessary software costs and
complexity of these methods are less than the rule or model based ones. (In the case of the
neural nets it is true only if the net is not trained on-line.) .

i.c
The KSC allows the different modules to run independently, to cooperate as agents
or to control each other. The third case means that one module is started by another one

og
because either the second one uses the results of the first one or the inference of the first
one led to the need of the second module.
Generally the resources of the KSC can use more clients (controllers)
nty
simultaneously. It leads to a cost effective AI solution, because one costly AI tool can
solve all the intelligent problems in a distributed environment. The overhead of the KSC
(network connection, one more computer, some delay etc.) is much less comparing to the
advantages (adaptive control tool licensing, less computing power in the
de

clients/controllers, one server module may used by more clients etc.).
Using the KSC together with the component based software technology (e.g.
stu

CORBA) gives a very adaptive software frame to solve complex problems.
In the Fig. 1 a CNC with an embedded PLC controls a machine tool. The modules
of both controllers are open and some of them are also clients of a knowledge server
(KSC). It means that these modules can run special adaptive control Intelligent Open
w.

CNC System Based on the Knowledge Server Concept methods during their work that is
an independent service is implemented in the KSC.
ww



om
i.c
og
nty
6. PROTOTYPE INTELLIGENT CNC BASED ON KSC:

In an early prototype of the intelligent open CNC that is using KSC, Adaptive
de

control system was implemented
An advance axis tester is put on the top of this. Axis Test
module handles all the tests but it gets the necessary position and velocity values from a
stu

knowledge based general tester running as an application on the KSC. The KB tester
determines some goal positions and motion speeds, that the Axis Test module executes
with the axis using jog commands. The results (execution time, tuning in errors etc.) are
w.

sent to the KSC that analyses and qualifies the axis. In the prototype the modules are built
in CORBA, the controller and the HMI is programmed in Java.
ww



7. CONCLUSIONS:
The controllable parameters in machining by using micro controller i.e. adaptive

om
control system are cutting force, torque, vibrations, feed and depth of cut. The
controllable parameters in machining by using artificial intelligenceModel based
on-line path generation
• Automatic tool selection

i.c
• Technological based settings of the operational parameters
• Automatic compensation of machine limits

og
• Automatic back-step strategies
• Detection and compensation of geometrical deflection
• On-line selection of control algorithms
nty
• Intelligent co-operation with other devices to solve problems
together
• Detection and correction of tool wear and breakage
•
de

Automatic handling of rejected work pieces
• Detection and managing of emerging situations of the machine tool
By introducing an interface between micro controller of adaptive control system
stu

and the data base of artificial intelligence we can control all the above parameters
by communicating with knowledge server. The features of KSC were discussed
and an early prototype was introduced.
w.

References:
ww

Adaptive control system -------- Bernard wplrow,Samuel D.stearns
Manufactureing Engg and technology by Serope Kalpak Jain, Steven R. schmid
Computer integrated manufacturing---------------- James A Regh, Henry w scrab
Manufacturing system Engg -------------------- katsundo Hitoni



om
i.c
og
nty
de
stu
w.
ww


Adaptive control-systems-paper-presentation-100115092600-phpapp01 (1)

More Related Content

Similar to Adaptive control-systems-paper-presentation-100115092600-phpapp01 (1) (20)

Recently uploaded (20)

Adaptive control-systems-paper-presentation-100115092600-phpapp01 (1)