Design of an Automatic Control Algorithm for Energy-Efficient ...
Design of an Automatic Control Algorithm for Energy-Efficient ...
Design of an Automatic Control Algorithm for Energy-Efficient ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
8 System <strong>an</strong>d functionality integration 91<br />
With the presented algorithm, <strong>of</strong> course m<strong>an</strong>y more settings <strong>an</strong>d modes would be<br />
possible. Through ch<strong>an</strong>ging weights the system c<strong>an</strong> be adapted online <strong>an</strong>d a personal<br />
trade-<strong>of</strong>f between the objectives could be set. However, this would lead to a very complex<br />
control, which is not desired. New ideas <strong>for</strong> user control c<strong>an</strong> be realised in a simple way,<br />
though.<br />
Finally, the car should have the possibility to give m<strong>an</strong>ual control to the passen-<br />
ger. Some people like to influence the climatisation directly <strong>an</strong>d do not w<strong>an</strong>t to give <strong>an</strong><br />
algorithm full control (although this is what a designer <strong>of</strong> such a program w<strong>an</strong>ts to make<br />
superfluous). Additionally it is useful in case <strong>of</strong> a hardware (sensor) or s<strong>of</strong>tware failure.<br />
There<strong>for</strong>e, m<strong>an</strong>ual control has to be made available in parallel to the automatic control.<br />
8.5 System simulation<br />
For the system simulation the differential equations derived in Chapter 3 c<strong>an</strong> be directly<br />
used in the given integration scheme from Section 6.3.4. Starting values are obtained by<br />
the sensors or the estimation (��� �����������).<br />
A good choice <strong>for</strong> a predicted horizon is in the r<strong>an</strong>ge <strong>of</strong> the system time const<strong>an</strong>ts<br />
(estimated in 3.1.5) in order to cover the system behaviour. To keep the computation time<br />
small, the choice <strong>of</strong> very few steps is proposed here. This makes the prediction inaccurate<br />
<strong>for</strong> bigger ch<strong>an</strong>ges, but this is supposed to be acceptable since accuracy is only required<br />
when reaching a steady state value with little ch<strong>an</strong>ges in the system states.<br />
8.6 Optimiser algorithm parameters<br />
The optimiser algorithm needs to be parameterised in order to be adapted to the problem.<br />
The parameters affect the computation time as well as the solution quality.<br />
8.6.1 Population size<br />
The population number consists <strong>of</strong> three figures:<br />
���� � ����� � ������ � �������