Tài liệu A Computer-Aided ppt

© 2001 by CRC Press LLC

4

A Computer-Aided and
Integrated Machining

System

4.1 Design Philosophy
4.2 User Interface
4.3 Feature Extraction
4.4 The Feature Recognizer

AutoCad Data • Processor • Analyzer • Initial Feature
Builder • Machining Feature Builder

4.5 The Knowledge Base
4.6 Factual Knowledge

Data Content

4.7 Procedural Knowledge

Selection of Operations • Sequencing of
Operations • Selection of Tools • Selection of Machining
Conditions • Generation of Tool Paths

4.8 Concluding Remarks
In order to arrive at a suitable solution, be it for process planning or part programming, the pressing

issues faced in an integrated manufacturing system are in the following.
1. Interpretation of CAD data for machining-based operations.
2. System integration in the activities involved in process planning, machinability data selection, and
tool path generation.
3. Update information arising from progress in manufacturing technology.
4. Decision making relating to manufacturing tasks.
5. Optimization strategy suitable for CNC processing capabilities.
The concept of an integrated machining system for the generation of production plans is introduced
in this work. The development of methodologies to address the above issues is presented. This chapter
expounds on the work reported by Yeo, Rahman, and Wong [1991] and Yeo [1995].

4.1 Design Philosophy

In this section, the concept of an integrated knowledge-based machining system is described. A software
tool, GOLDWORKS

TM

[Gold Hill, 1989] with COMMON LISP as the programming platform, is used
for the development work. The software tool requires Microsoft Windows

®

in a PC which can be feasibly
deployed in a wide range of industries. The system is aimed to provide a vital step to a totally integrated

Swee-Hock Yeo

Nanyang Technological University

© 2001 by CRC Press LLC

manufacturing environment. The integrated machining system (IMS) begins with the interpretation of a
part drawing and transformation of the geometrical data into a frame structure of machining features. The
system then assigns appropriate operations to all features identified. Automatic tool selection of holders and
inserts follows before planning the entire operation sequence. For each operation, the machining conditions
are determined. Finally, tool paths are generated for a selected CNC lathe.
The assumption made in developing the proposed IMS is that component set-ups with suitable work-
holding parameters are provided so that the solution generated by the IMS can be implemented. Proper set-
up involves spatial reasoning, and it is a difficult task that an experienced machinist learns during an appren-
ticeship. Varying methods of solution exist from simple rules-of-thumb, such as threshold ratio between length
and diameter and analytical methods such as that of explained by Hinduja and Huang [1989]. The problem
of specifying the appropriate set-up is further complicated by the variety of work-holding devices which include
face plates and fixtures, mandrels, jaw-type chucks, step chucks, collets, and magnetic and vacuum chucks.
The architecture of the system that has been developed in this research is illustrated in Figure 4.1. The
system is comprised of a user-interface for interaction, a feature recognizer, and a knowledge base that
contains facts and rules. Figure 4.2 shows the various types of approaches/strategies that have been used
for the machining planning tasks. It also outlines the sections of this chapter for each of these tasks.
The inferential strategy used for the decision-making process of the IMS is based on the goal directed
forward chaining technique (which integrates both forward and backward chaining techniques) [Yeo, 1995].

FIGURE 4.1

Architecture of the integrated machining system.

© 2001 by CRC Press LLC

For a given goal, the chaining method uses rule sets to cluster forward-rules together, and each rule set is
activated when an enabling pattern is matched and queried. The control structure consists of eight stages where
each stage is made up to a rule set as depicted in the following with a brief description in the right column.

The inferencing mechanism is fired by backward chaining via the THEN part of the control structure
with the pattern named RUN-STATE SELECTION COMPLETED. Each rule set is activated sequentially
when its enabling pattern (e.g., RUN-STATE FEATURE-RECOGNITION DONE) is matched and the
inference mechanism performs forward chaining of all the rules related to the rule set. The benefits of
having such methods are having one group of rules fired before another group and improving code efficiency
of the matching process in which only appropriate rules for a particular state are used.
Outputs obtained from the IMS are composed of two main parts namely (see Figure 4.1).
1. A process sheet for each set-up, which is a set of instructions consisting of the following.
• Enumerated operations with lists of machining features.
• Cutting tools (holders, inserts, drills, etc.) for the operations planned.
• Positions of cutting tool mounted on the turret.
2. A CNC part program consisting of machine codes.

FIGURE 4.2

Decisions modes in the integrated machining system.

(DEFINE-RULE SELECTION-CONTROL ; production rule with a series of enabling
patterns
(RUN-STATE FEATURE-RECOGNITION DONE) ; define machinable features, stock size, etc.
(RUN-STATE INTERNAL-OPERATION DONE) ; selection of internal operations
(RUN-STATE EXTERNAL-OPERATION DONE) ; selection of external operations
(RUN-STATE OPERATION-SEQUENCING DONE) ; assignment of operation sequence
(RUN-STATE TOOLHOLDER-SELECTION DONE) ; selection of toolholders
(RUN-STATE INSERT-SELECTION DONE) ; selection of inserts
(RUN-STATE MACHINING-CONDITION DONE) ; selection of machining conditions
(RUN-STATE TOOL-PATH-GENERATION DONE) ; generation of tool paths
THEN
(RUN-STATE SELECTION COMPLETED) ; query of the attempt to fire the rule

© 2001 by CRC Press LLC

4.2 User Interface

The purpose of this user interface is to provide interaction between the user and the system, data input,
and data modification during the machining planning process. The user interface is driven by means of
a series of menu-driven facilities and graphic images. These include the following.
1. A screen layout with a row of commands for pull-down menus.
2. A window with a set of buttons for performing machining planning tasks.
3. A window for schematic display of a task.
4. A status bar window to monitor the progress of the machining planning tasks.
5. A window for the textual explanation of the problem solving procedures.
6. Pop-up menus for the data input.
In addition, there is a help facility for obtaining information regarding tooling, work materials, etc. Error
trap facilities are also incorporated in the system, for example, input of a null value or an erroneous value.

4.3 Feature Extraction

Since this research deals with 2-D operations for producing rotationally symmetric parts on lathes, a
plan view of the upper half cross-section of the part drawing of the workpiece will suffice. The form of
the profile is restricted to straight lines and arcs. The feature recognizer requires access to part descriptions
stored in a CAD database. The approach that has been adopted in this work is to develop a module that
involves inductive extraction of the geometrical data of a part created in a commercial CAD system. Its
DXF (in AutoCAD®) facility provides the output format required by the IMS. Before any operation
planning activity can take place, a set of features must be formulated for abstraction. Since this work is
directed at metal removal processes, machinable features are used. A machinable feature is any geometric
surface or combination of geometric surfaces which form a volume that can be shaped by cutting tool
operations. The term “form features” is sometimes used.
The necessary information about each feature is organized by means of a frame representation. This
knowledge representation provides modularity and convenient accessibility of the knowledge base.

4.4 The Feature Recognizer

The feature recognizer module has five steps which are shown in Figure 4.3 as described:

AutoCad Data

A section view of the top half of a part is created using AutoCAD®. Line-types of each entity and their
significance have been defined as shown in Figure 4.4 so that the CAD data can be used effectively in the next
step. All the entities representing the model and the stock size are exported using the AutoCad DXF format.

Processor

The Processor accepts a DXF data file and processes it into a form acceptable to the analyzer. The
COMMON LISP list-data structure has been used to store the DXF data into a temporary file. An example
of the temporary file is as follows:
(0 SECTION 2 ENTITIES
0 POINT 8....
0 LINE 8....
0 ARC 8....
....
0 ENDSEC 0 EOF)

© 2001 by CRC Press LLC

The temporary list file is then processed further by removing integer notations and group entities into
a more refined list data file as follows:
Each entity type and geometrical data is contained in a sub list.

Analyzer

The analyzer determines the nature of each entity and assigns to each entity a primitive feature which
defines functionality. The primitive features include vertical_line, horizontal_line, slant_line,
threaded_line, convex_up, convex_down, concave_up, and concave_down. The various line types are

FIGURE 4.3

Structure of the feature recognizer.

((LINE (178.2 142.2) (221.3 165.2)) ; sub list for each
.... ; entity type
(ARC (141.7 151.7) 5.0 (180.0 270.0))
)

© 2001 by CRC Press LLC

easily understood. The various arc types are illustrated in Figure 4.5. Their significance is used for tool
selection. For the example used in (“convex up”), the sub list resulting from this analysis gives the
primitive list data as follows:
Each sub list describes the types of primitive. The entire primitive list data is sorted in order, starting
from the left-most primitive of a part.

Initial Feature Builder

Using a frame lattice structure shown in Figure 4.6, the initial feature builder converts the primitive list
data into embryonic objects (i.e., instances of frames) which are exclusively geometric. The machining
features of a component are represented in the knowledge base in the form of frames with hierarchical

FIGURE 4.4

Significance of the line type.

FIGURE 4.5

Element library of arc types.

((LINE (178.2 142.2) (221.3 165.2) (PRIMITIVE SLANT_LINE))
....
(ARC (141.7 151.7) 5.0 (180.0 270.0) (PRIMITIVE CONVEX_UP))
)

© 2001 by CRC Press LLC

relationships. The lattice structure consists of eight types of frame used for a rotationally symmetric
part, which are: plane, horizontal, vertical, taper, arc, chamfer, groove, and thread. Seven of these types
of frame may have associated instances. Two slots in the feature frame are used to provide connectivity
to adjacent features. The plane-type frame groups three other frames which are all concerned with
straight lines.
An example of a taper feature instance in the language syntax form is shown in Figure 4.7. The EXT6
instance is related to the taper frame which is a child frame of the plane frame. Beside the representation
of feature types, each instance is linked to its adjacent features (i.e., EXT5 and EXT7), thus forming a
linked list. For convenient identification, each instance is uniquely named and enumerated, for example
EXT and INT imply external and internal feature-types respectively, and the numbers appended to them
are arranged in order starting from the left-most feature of the part.

Machining Feature Builder

The machining feature builder makes a complete set of all the machinable features of a part for the
knowledge base. Production rules are used in the pattern matching process. The rules include recognition
of grooves, chamfers, and threads. Surface roughness and geometric tolerances can be added manually

to the attributes of a feature as appropriate.
While the recognition of chamfers and threads are easily implemented, the recognition of grooves
requires a more detailed geometrical treatment. A groove (or recess) in a machining profile is generally

FIGURE 4.6

Frame lattice structure of the machining features for a rotationally symmetric part.

© 2001 by CRC Press LLC

described as being confined by two adjacent boundaries. This definition is very broad and might lead to
inefficient automatic tool selection. Grooves may be required in a wide variety of forms such as: circlip
grooving, O-ring grooving, face grooving, deep grooving, wide grooving, and undercutting.
In this work, the recognition of groove is tool-oriented. There are various possibilities for machining
a groove feature consisting of a horizontal element bounded by two vertical elements as illustrated in
Figure 4.8. The use of two tools, that is, two longitudinal tools (Figure 4.8a) or a longitudinal tool and
a grooving tool (Figure 4.8b), requires two tool positions on the turret; thus there is an economic
implication. One grooving tool (Figure 4.8c) could be used to solve the problem of limited turret capacity.
It may not be economical if a large number of passes is required, i.e., for a wide groove.
To produce a rectangular groove feature, i.e., both the vertical elements parallel to the

X

-axis and the
bottom of the groove parallel to the

Z

-axis, the external tool must have an approach angle of more than
90°, and the largest possible trailing edge angle. An insert shape of 35° included angle would be suitable.

Figure 4.9 shows the best choice among the standard tools available for external operations. Using a 25 mm
square shank toolholder, the groove depth (l) should not be more than 42 mm and width (

w

) at least
64 mm, depending on the clearance provided, hence, the critical

w/l

ratio is equal to 1.524.

FIGURE 4.7

An example of a taper feature instance.

© 2001 by CRC Press LLC

As compared with this tool-type, a standard 25 mm square shank grooving tool has a

w/l

ratio of 1.63
(see Figure 4.10). Though the

w/l

ratio for grooving tools is marginally less than that of the external tool,
the size of the groove can be as deep as 16 mm with 26.2 mm width. The former tool cannot produce
this shape.

FIGURE 4.8

Three possibilities for machining a groove feature.

© 2001 by CRC Press LLC

The machining feature builder is designed in such a manner that modification to the knowledge base
can be done easily. Thus, the production rule for a groove can be extended to provide for a wide groove,
a narrow groove, and a deep groove. Other form tools for cutting specific grooves or recesses are not
considered in the present work. An example of a rule to recognize a groove-feature is as follows:
Using the example given in Figure 4.7, Features EXT2, EXT3, and EXT4 satisfy the groove-feature rule and
are combined into a groove feature; thus, a frame-based approach to formulate a lattice structure of machining
features in a rotationally symmetric part has provided an efficient means to represent a generic model. The
organization of the lattice structure is suitable for efficient pattern matching of rules. The methodology
has been used for a wide range of machining features in a part, and provides the flexibility to change feature
definitions with minimal effort.

FIGURE 4.9

Geometrical constraint of an external cutting tool to machining a recess.

FIGURE 4.10

External grooving tool.

IF
surface, F1 is adjacent to horizontal surface, F2, and
horizontal surface F2 is adjacent to surface, F3,
F1 & F3 are 90° to F2, and

F2 with width of less than 26.2 mm
and
F1 with depth of less than 16 mm
THEN
surfaces F1, F2, F3 form a feature GROOVE

© 2001 by CRC Press LLC

4.5 The Knowledge Base

The types of knowledge acquired are a combination of factual (or declarative) knowledge and procedural
knowledge, which involves selection of: operations, operation sequencing, selection of tools, selection of
machining conditions, and generation of tool paths.

4.6 Factual Knowledge

A sizable amount of data is required for machining decision making. Specifications of work materials,
toolholders, inserts, etc., must be stored in the files of the data base management system, with maintenance
of these files being carried out as necessary.
As indicated in Figure 4.11 an interface between the knowledge base and the data base files must be
employed for the retrieval of relevant data. External data base files, where data searching must be done
sequentially on the index key until the search key field is matched, must be accessed by the interfacewthis
way, besides improving the computational time, the data base management system can also be used to
serve other organizational functions, such as tool management which is important to improve overall
productivity. The crucial issues, namely system integration, ease of data accessibility and system flexibility
are thus taken into account.

Data Content

The data base files for recommending machining data required by the knowledge-based system have been

described by Yeo, Wong, and Rahman [1991]. In order to search and retrieve data from the work material data
base files, the dbase-action frame structure shown in Figure 4.12 serves as an interface to the external files.

FIGURE 4.11

Data acquisition system for the knowledge base.

© 2001 by CRC Press LLC

For example, a data base file is opened by using the function:
The variable interface is an instance of dbase-action frame. The action slot-name of ‘:open-file’ simply opens
a data base file when the go slot-name is ‘:yes.’ A further action named ‘:find-record’ searches the file until
the search-key slot-name is matched and the result (i.e., record) is asserted in the value slot-name.
The part description of an instance named THE-PART-INFO is shown in Figure 4.13. The assertion
of the stock size is obtained from the DXF file. The material specifications are retrieved and asserted to
the relevant slots using the above interface procedure. External data accessing is done quickly with ease
even with a large data file size. If numerous data queries are made on the same file, for example the
cutting tool data files used for checking tool availability and tool selection, it would be convenient to
convert the data into a frame system. A particular machine tool with its attributes is represented as an
instance of the machine tool frame THE-MACHINE-INFO in Figure 4.14.
Data base structure for cutting tools can amount to more than 20,000 combinations of tool modules
[Evershiem, Jacobs, and Wienand, 1987]. It is important to provide efficiently in the data base structure
for the following.
1. Ease of maintenance.
2. Complete and concise attributes of tools.
3. Linkages between holders, inserts, etc.
A tool database structure consisting of 21 data fields has been formalized and is used for external and
internal holders as well as for drills. Examples of a tool file containing three types of tool are shown in Table 4.1.

FIGURE 4.12

Frame structure of a dbase-action.

(Defun open dbase (interface)
(setf (slot value interface ‘action) :open file)
(setf (slot value interface ‘go) :yes))

© 2001 by CRC Press LLC

The purpose of each data field is as follows.
1. Identification code.



This field is used to designate a unique identification for a tool holder or a drill.
2. Insert clamping method.



This field is used for four types of clamping systems which are designated
in the ISO standard, i.e., C, M, P, and S types. The C type provides a positive rake which is usually
preferred for finishing operations, while the M type provides rigidity and permits use of double
sided inserts. The P type offers unobstructed chip flow in addition to the advantages for M type.
The S type is used for finishing operations and provides a larger variety of insert shapes. The C type
is largely used in ceramic inserts.
3. Coupling (shank) size. This field is used for the specification of shank height, shank width, and
tool length, which affect the stiffness of the tool. Its size is restricted by the turret tooling capability.
4. Hole diameter.

This field is used to ensure that a boring bar is able to function without colliding
into the workpiece. For example, a 25 mm bar diameter should have a minimum hole diameter
of 32 mm when the bar is placed at the centre of rotation [Sandvik Coromant, 1990]. In the case
of drilling, the data field refers to the drill diameter.
5. Tool function.



This field is used for the classification of operations (sometimes referred to as process
classification). The hand of tool (i.e., left, right, or neutral) specified in the ISO standard is insuf-
ficient to describe the full capability of a tool such as that illustrated in Figure 4.15. The right-hand
tool, PDJNR, shown in Figure 4.15a is used for profiling while PCLNR, shown in Figure 4.15b is
used for longitudinal turning and out-facing, as well as for in-facing.

FIGURE 4.13

Part description of an instance named
THE-PART-INFO.