An ANN based Cruise Control Method for a Body-Caudal Fin Propulsion System for Autonomous Underwater Vehicle

Abstract

Bio-mimetic propulsion systems are gaining importance than conventional propulsion systems for the endurance and efficiency. For higher speeds the Body-Caudal Fin (BCF) propulsion systems are preferred than Median-Paired Fin (MPF) propulsion systems among oscillating fin propulsion systems to be used in an Autonomous Underwater Vehicles. The natural swimmers are able to cruise in the water by adopting the parameters like frequency and amplitude of oscillations based on the information learnt by their brain over the time. A cruise control method for BCF propulsion employing a rigid fin is proposed by mimicking the creature’s brain with an Artificial Neural Network (ANN). The method will be to train a Feed-forward Back-propagation network with performance parameters like swim velocity, average thrust, propulsive efficiency etc. evaluated through numerical simulations as the inputs and frequency and amplitude as the target data. The swimming conditions are selected within the range of 0.15 to 0.4 of the Strouhal number. Ninety data sets are prepared for different frequencies and amplitudes. The trained Neural Net predicted the oscillating parameters with 90% accuracy or higher.