00191413-01.pdf - 第324页
11 Production Tools / Optimization User Manual Line Computer UNIX 11.2 Concept Used Software Version 501.xx 01/99 Issue 11 - 4 11.2 Concept Used To be able to s olve the optim ization pr oblem in the f irst plac e, it wa…
User Manual Line Computer UNIX 11 Production Tools / Optimization
Software Version 501.xx 01/99 Issue 11.1 Fundamental Information
11 - 3
Mathematically, the optimization problem belongs to the class of NP (NP - non-deterministic polynominal)
problems, i.e. the problem cannot be solved by deterministic means in a finite time.
To get an idea of the magnitude of the optimization problem, some maximum figures that must be reckoned
with in extreme cases are listed below:
100 PCBs per optimization job
3000 placement positions per PCB
1000 different components
6 placement stations per line
5 different station types and system components
33 different feeders and waffle-pack trays
5 different placement heads
2 different head types
6 different auxiliary units
The objective of set-up optimization is as follows:
11.1.1 Functionality Extension
Optimization has been extended by the „Setup clustering“ functionality.
Unlike before, when the optimization run used to be aborted together with the release of the error message
„Insufficient line capacity“ more than one setup is now being calculated. To this end, the PCBs are combined
into clusters that can be assembled with the same setup. Not only the aforementioned clustering is performed
but also the changeover operations taking place between the setups are taken into account.
In this way, the entire machine running time (= production time + non-productive time) of an SMT line can now
be optimized for the first time.
"Setup clustering" is available as an additional option.
The minimization of both the total of all production times of all scheduled PCBs and the total of
the changeover times between setups of a line of any configuration.
11 Production Tools / Optimization User Manual Line Computer UNIX
11.2 Concept Used Software Version 501.xx 01/99 Issue
11 - 4
11.2 Concept Used
To be able to solve the optimization problem in the first place, it was subdivided into several tasks. The individual
tasks are shown in Fig. 11.2.1 in the form of different layers. The individual layers together with their specific
tasks are described in the following.
Clustering
The foremost task of Clustering is to determine whether more than one setup is required. If this is the case, the
PCBs are grouped into clusters each of which can be produced with one setup. The underlying objective is to
create the setups in such a way that the changeover times from one setup to the next are minimal. This objective
can be computed according to two different setup methods (i.e. sequence or mean value) (see section 11.4.3.1).
Balancing
The task of Balancing is to distribute the components among the placement heads of the line in such a way
that, on the one hand, the available capacity in the feeder part is not exceeded and, on the other hand, the
stations of the line are utilized evenly for each PCB. It must be kept in mind when so doing that given components
may be part of several PCBs. By the way, the need for optimization results from the fact that the available space
is limited. If there was plenty of space in the individual feeder parts to supply all components at all placement
heads simultaneously, there would be no need for Set-Up Optimization. The problem could then be reduced to
the uniform distribution of the placement positions.
Track Allocation
The task of Track Allocation is the optimum setup of the components allocated to the respective feeder part.
The assumptions made in the Balancing layer must now be fulfilled in this layer to the greatest possible extent.
Placement Sequence
The task of Placement Sequence is to allocate the placement positions to the placement heads of the line, on
the one hand, and to calculate a placement sequence for each head, on the other. A constant setup is taken
as the basis, i.e. the setup is adopted from the higher-level layers without any further changes.
Fig. 11.2.1 Layer Model of the Optimization Problem
Balancing
Which component should by placed by which placement head of the line?
Clustering
Which PCB is to be produced with which setup?
Track Allocation
At which location within the feeder part is the component to be set up?
Placement Sequence
Which placement position should be populated by which head?
In which sequence are the placement positions to be populated?
User Manual Line Computer UNIX 11 Production Tools / Optimization
Software Version 501.xx 01/99 Issue 11.2 Concept Used
11 - 5
The layers have intentionally been drawn one inside the other since interactions between the layers also occur
in the actual optimization problem. Even though the layers are dependent upon one another they should be
solved independently of one another. The optimization problem is thus rendered resolvable, although an error
will be produced. In order to keep the error as small as possible, the simplified assumptions of the higher-level
layers must be fulfilled by the lower layers, as far as possible. The individual layers must be so coordinated with
respect to one another that the error is of minor significance for the optimization result.
In Set-Up Optimization the tasks of the Clustering, Balancing and Track Allocation layers are processed. The
Placement Sequence layer is contained in the postprocessor for PCBs and is always then calculated when a
producibility check is activated or a job scheduled. The Placement Sequence cannot be calculated any earlier
since the current state of the data is needed as the basis. In particular, loading a new PCB causes a lot of
changes at the data base. Such changes can affect the placement sequence as well. This is the reason why a
placement sequence must not be calculated prior to this point.
When using the optimization option, the operator first determines the types and the number of the PCBs to be
assembled. Following a plausibility check, optimization starts. As a result of optimization, the optimization
processor then creates one or several setups (see section 11.3.2) which are stored in the master data storage
under the path „Anlagen/<line>.ak /<subline>.va/*. In addition, the lot file is modified. For each job (PCB type
and nominal lot size) contained in the lot file, the name of the line, the subline and the setup are entered. After
optimization has been completed successfully, the jobs contained in the modified lot file can be merged with
the job list of Job Control and subsequently be directly scheduled.
In order to retain the information between PCB and setup, the job will not be deleted from the job list following
its completion.
NOTE
No setup will be created if it were to contain a PCB for whose components and feeders there would not be
sufficient capacity on the line.