Vortex induced vibration (VIV) occurs when vortex shedding frequency falls close to the natural frequency of a structure. Investigation on VIV is of great value in disaster mitigation, energy extraction and other applications. Following recent development in machine learning on VIV in laminar flow, this study extends it to the turbulent region by employing the state-of-the-art parameterised Navier-Stokes equations based physics informed neural network (PNS-PINN). Turbulent flow with Reynolds number Re = 10 4 , passing through a cylinder undergoing VIV motion, was considered as an example. Within the PNS-PINN, an artificial viscosity ν t is introduced in the Navier-Stokes equations and treated as a hidden output variable. A recently developed Navier-Stokes equations based PINN, termed NSFnets (Jin et al., J COMPUT PHYS 426: 109951, 2021), was also considered for comparison. Training datasets of scattered velocity and dye trace concentration snapshots, from computational fluid dynamics (CFD) simulations, were prepared for both the PNS-PINN and NSFnets. Results show that, compared with the NSFnets, the PNS-PINN is more effective in inferring and reconstructing turbulent flow under VIV circumstances. The PNS-PINN also shows the capability to deal with unsteady and multi-scale flows in VIV. Inspired by the Reynolds-Averaged Navier-Stokes (RANS) formulation, by implanting an additional artificial viscosity, the PNS-PINN is free of complex turbulence model closure, thus may be applied for more general and complex turbulent flows. Keywords Machine learning • Vortex induced vibration (VIV) • Turbulent flow • Physics informed neural network (PINN) • Reynolds-Averaged Navier-Stokes (RANS)