<p>Based on experimental data, a data error estimation model is developed for an existing neural network time dependent hydrodynamic force model in order to allow dynamical analysis and controller systems for the dynamic forces on a circular cylinder which is forced to oscillate in two-degree-of freedom in free stream under the occurrence of Vortex-Induced-Vibrations (VIV). This model is used to developed future control models to improve energy harvesting from VIV-based systems. Experimental data series were recorded with extensive amount of unique automated experiments which includes force measurements of the fluid forces on the rigid cylinder which is forced to oscillate in-line and cross-flow in free stream at Re=7620 with nearly 10,000 force measurements experiment. The experiments were conducted in a fully automated towing tank with the motion parameters of cross-flow motion amplitude, in-line motion amplitude, reduced velocity and the phase between in-line and cross-flow motion. Previously, dynamic model of appropriate complexity was created in order to analyze lift and drag forces from force database. Position, velocity signals were used as input to the dynamic model. A feed forward neural network was trained using the force database in order to develop time dependent models of forces on the cylinder where it is forced oscillate by prescribed sinusoidal motion. The time series error between the measured and feed forward Artificial Neural Network (ANN) model was found for lift and drag force time history. Artificial Neural Network model was built for estimating time series of lift and drag force because the theory suggests stationary and ergodic signals in order to build autoregressive error (AR) predictor. Autoregressive (AR) model is a representation of a type of random process such that it is used to describe certain time-varying process. The autoregressive model (AR) shows that the output variable depends linearly on its own previous values in order to make prediction. Therefore, Autoregressive (AR) error predictor is used from existing error results between measured and artificial neural network model in order to make error predictor. In this study, autoregressive error predictor is developed from existing neural network time dependent model of forces on the cylinder for given kinematic conditions by using autoregressive model (AR). This autoregressive (AR) error predictor from measured signal and artificial neural network model will help us to improve the model. Using this autoregressive (AR) error predictors help us to predict errors to correct neural network model. This data driven model is very important in order to understand better the physics and dynamics involved in VIV. It will contribute greatly toward energy harvesting control system developments. &nbsp;</p>

Error Analysis of Models for the Forces on a Cylinder Undergoing 2-Dof Prescribed Motion in a Stream

Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2021/volume/85543

ASME 2021 International Mechanical Engineering Congress and Exposition (IMECE2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.