Model-Based Self-Tuning PI Control of Bolt-Nut Tightening for Wind Turbine Bearing Assembly

摘要

One of the core steps of the assembly of wind turbines is the assembly of the bearings on the wind turbine hub. The hub can contain up to 128 bolt connections to install the bearing blades: nuts need to be precisely tightened to ensure a uniformly distributed clamping force as well as avoiding assembly errors, e.g. nut misalignments. The bolt-nut connection is a non-linear system with uncertainties making it difficult to design a numerical model and PI Gains. This paper presents a novel two-stage Proportional-Integral (PI) controller with assembly error detection capability for bolt tightening process. It is based on the combination of a numerical model (offline training) and a genetic algorithm (GA) for online training on the physical bolt system. Since the model does not include all non-linearity and uncertainties of the physical plant (here the bolt-nut connection), it is used at first to estimate the range of the PI values, followed by a fine tuning of the values online by the GA.