Establishment of a flow-induced vibration power database based on deep neural network machine learning method

Abstract

In this paper, a machine learning method based on Deep Neural Network (DNN) is used to establish a predictive model for flow-induced vibration (FIV) power output, and the power database is constructed based on this model. The results reveal that the power database can quickly and efficiently obtain power values for any parameter combination, significantly improving the efficiency of studying the power response of flow-induced vibration. Its accurate prediction ability, low cost and time consumption have made the database a useful tool. Furthermore, the power database can provide power values for various parameter combinations to comprehensively analyze the impact of system parameter combinations on power. The screening results indicate that the optimal power output for the single oscillator flow-induced vibration energy harvesting is 22.71W and the optimal power for the dual oscillator FIV energy harvesting is 36.38W. A power database of single and dual oscillators flow-induced vibration harness energy provides a novel method for the study of flow-induced vibration power generation.