Open AccessDevelopment and embedded deployment of a virtual load sensor for wind turbine gearboxes
Author(s) -
Jelle Bosmans,
Yon Vanommeslaeghe,
Luk Geens,
Jakob Fiszer,
J. James R. Croes,
Matteo Kirchner,
Joachim Denil,
Paul De Meulenaere,
Wim Desmet
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1618/2/022011
Subject(s) - downtime , turbine , software deployment , strain gauge , computer science , real time computing , virtual machine , simulation , automotive engineering , engineering , structural engineering , mechanical engineering , operating system
The loads to which a wind turbine gearbox is subjected during its lifetime can be a valuable source of information to decrease maintenance cost and downtime through enhanced monitoring, control and design. However, this load information is difficult to acquire since suitable direct load sensors are intrusive and expensive. Therefore, this paper focuses on indirect load measurement through a virtual sensing algorithm. The resulting virtual load sensor estimates the incoming load on the low speed planetary stage of the gearbox by combining strain measurements on the external surface of the ring gear with a physics-based model. The algorithm is deployed for real-time execution on low-cost embedded hardware to make a cost-effective load sensor. The effect of the configuration parameters of the virtual load sensor on the execution time and memory usage is examined in order to verify which configurations can be deployed. Since these configuration parameters also affect the estimation accuracy, the design of the virtual load sensor is tackled as a co-design problem. The resulting virtual load sensor, which is deployable for real-time execution, achieves an RMS estimation error of 0.6% in a numerical validation, using 4 strain gauges on the ring gear.