Effectiveness investigation of the RBF-neural network applied to modelling of the ship propulsion systems with controllable pitch propeller (CPP)

Abstract

Using the controllable pitch propellers (CPP) on ships introduces, apart from the rotational propeller speed N an additional degree ofb freedom: the pitch ratio P/D of the propeller, increasing the setting space of the propulsion system but also imposing the necessity of the optimum setting determination. To realize the aim, as accurate as possible characteristics of the propulsion system must be at disposal. The characteristics can be determined with the use of both the conventional mathematics and artificial intelligence methods such as neural networks or fuzzy logic. The paper presents a method of neural modelling of the propulsion system and results of the experiments which makes evaluation of the network effectiveness posssible.