The method of underwater object identification using multi-layer perception

An algorithm for detection/identification of underwater objects is proposed. The algorithm is based upon the classification ability of a simple multi-layer neural network, also called a Perceptron. A signal recorded by a hydrophone and preprocessed by a computer is supplied to the ANN, which classifies it according to the possessed information encoded within its structure and an array of weighs. Analysis of effectiveness is conducted depending on the variables pertaining to the neural network.