Adaptive Surface Ship-Wake Detection Based on Improved One-Class Support Vector Machine

It is difficult to collect bubble-wake signals from different ocean environments caused by various types of ship. One-class support vector machine (OCSVM) can make decision based on incomplete information. This paper found an OCSVM detection model which use only reflected signals without a bubble wake to detect the surface ship-wake. In order to improve the training efficiency, a training algorithm based on Sequential Minimal Optimization (SMO) was introduced for OCSVM. Grid search method and Particle Swarm Optimization (PSO) algorithm are used to search optimal parameter. The simulation shows that the proposed detector can detect the ship weak well and it was robust with respect to noisy signals.