Modeling And Predicting Of Neural Network And Application In Watercraft Motion Modeling And Prediction

Conventional time series analyZing and predicting theories are on base of1inear model, and it is an effective means for linear systems. There are amountof nonlinear systems in practical projects, so it is very imPortan to studymodeling and predicting of nonlinear systeIns. When large watercrafts movein the sea,they are affected by ocean wave' sea breeze and other disdris, asa result they move at random and nonlinearly With six convertibilities, so itbrings on considerable difficu1t fOr predichon of watCrcrall motion eAner intheory or in practical prOject.This paPer discuss a modeling and predicting means for nonlinearsystems proceeding from nonlinear syStems modeling and Predicting theory,Whch is based on DRNN model. This means overcomes the fact that ARmodel is used only in linear systems, at the same time it connects itself withaPproximation theory symbolic statistics and conjugate gradient algorithm,and formulate a system of large watercrafts motion modeling and predictingWhich is based on DRNN model, and simulate it. As a conclusion the contentof this paPer is as follow;l. This paPer combines neural netWrk theory with nonlinear systemsmodeling and predicting theory and brings forward a means that is aPplicableto noulinear systems.2. Based on the scheme of modeling and predicting of neural network,the auhor select a neural network model Which is applicable to watercraftmotion short time modeling and predicting according to the character ofwatCrcraft historical data.3. The method of DRN'N modeling and predicting is given in this paPeLThe author works out the algorithm, gives the multi-step predictingalgorithm, and presefits the convergence theory of the algorithIn. In the endthe predicting model is used in watercraft motion modeling and predicting,and the auhor analyzes the result of simulating. The result indicates that thisap$j$Ifgx9@1%&itxmeans is reasonable and feasible and gains the satsfactory pmpose.