Generation of Fluid Field Around Tidal/Ocean Power Generator with Unidirectional Flow Exploiting Variable Auto Encoder

In this chapter, we aim to generate rich data useful for understanding the essence of natural phenomena. Rich data generation is required to predict/forecast situations involving renewable power generators interacting with natural events. Renewable energy is an essential factor in guaranteeing the sustainability of society. The Tsugaru Strait, in the northern region of Japan, is an area that has attracted attention for the utilization of tidal/ocean energy. We propose a tidal/ocean power generator utilizing a flaring flanged diffuser (FFD) to harness the power effectively. To obtain rich data around FFD, generative deep learning is focused. Variable auto encoder (VAE) is used to reconstruct, encode and decode, the fluid field data. The original data of the generation is obtained from the measurement data in the experiment. The architecture of the generative deep neural network is constructed by exploiting the decoder in the learned VAE. The reconstruction of the fluid field is performed, and the performance is evaluated.