• Kris Kristensen Riisager
  • Niels Christian Bender
4. term, Energy Engineering, Master (Master Programme)
ThisMaster’s Thesis investigates fault detection and
diagnosis (FDD) in a hydraulic pitch system by applying
artificial neural networks (ANN). A mathematical
model of the pitch system is developed
and the parametrisation of the model is conducted
based on experiments. The model is applied for
development and testing of several FDD schemes,
which are used to evaluate which experimental tests
are necessary to perform. A test rig including
the main components of the pitch system is constructed
in a manner that allows emulation of leakage
faults. The test rig allows for general evaluation
of the developed FDD scheme. The main conclusion
states that a model-based scheme with diagnosis
by a focused time-delay ANN shows theoretical
superior performance. This is tested on the test
rig to validate the scheme. The main results show
a decrease of estimation accuracy of leakages when
applied on the actual system, however, a stuck servo
valve and a pressure transducer failure are successfully
SpecialisationMechatronic Control Engineering
Publication date1 Jun 2016
Number of pages215
ID: 234543361