Prediction of Drag Reduction Parameters of microbubbles Based on Multi-hidden Layer BP Neural Network Algorithm

摘要

In this paper, aiming at the problem of the short endurance of current electric propulsion ships, the microbubble drag reduction system is improved. At present, there is no suitable algorithm to calculate the amount of air injection and the air injection equipment consumes too much energy. We use the multi-hidden layer BP neural network model to predict the optimal amount of micro-bubbles ejected and the optimal inclination Angle of the ship under different sailing states to achieve the purpose of dynamically controlling the amount of air injection and reducing the power consumption of the equipment. Through the analysis and integration of the optimum inclination angle and the amount of microbubble ejection of the experimental vessel under different speed and load conditions, the BP neural network was trained by generating learning samples. The multi-hidden layer BP neural network has the characteristics of fast convergence speed and high prediction accuracy. The experimental results show that the algorithm of multi-hidden layer BP neural network can predict the optimal bubble quantity and ship's camber Angle more accurately under the real-time sailing speed, and then significantly improve the effect of drag reduction of microbubble.