The Optimization of Geodetic Networks with Simulated Annealing
optimization of geodetic networks is one of the most difficult tasks
for geodesists. It needs much knowledge and experience to design an
accurate and reliable network which is easy to measure as well. The
process of designing networks has been poorly supported with computers
by now. They are mainly used for analyzing the design but are not yet
capable for making suggestions to improve the design.
shows how to optimize a network automatically. Network design is
treated as a combinatory problem. In case of a second order design the
optimizing procedure looks for the combination of measurements which
fulfills all quality criteria and needs the lowest effort for measuring
the network. Therefore the optimization procedure needs to know all
possible measurements in the network.
The thesis describes the
necessary quality criteria and the cost function for the optimization.
The cost function is not smooth and it has many suboptima. Therefore it
is not possible to use normal optimization techniques. Here Simulated
Annealing (SA) has been used, which can deal with many suboptima and
non smooth functions. With SA it is possible to find the optimum of the
cost function. To speed up the optimization and save computation time
it is not necessary to find the global optimum. It is enough to
approximate the optimum with a low suboptimum.
The algorithm for
optimization is tested with various examples. The tests show that an
optimization is possible. The algorithm proves to be useable in every
day geodetic network optimization.