Research On Intelligent Fault Diagnosis Method For Measurement System Of Dynamic Positioning Of Ships

Ocean is rich in resources,along with the development and exploration of ocean resources,the human beings started to focus on the resourceful ocean more and more passionate,the pace is getting faster and faster,making the marine equipment and marine engineering technology rapid developed.Dynamic Positioning System has become a necessary supporting system for ships operating at sea,and it is wildly used in field of the laying of submarine pipelines,fixed point operation and sea rescue.Measurement system is one of the most important parts of Dynamic Positioning System,which influences system's control accuracy and reliability directly.Measurement system intelligent diagnosis is deeply researched in this paper.The main contents are as follows:First of all,build the mathematic model,including the 3-DOF kinematics and dynamics models of ship and environmental load models of wind,wave,and current,so that it can truly reflect the response characteristics of ship under the influences of environmental disturbance.The parameters of the ship model used in the method validation are given,and the basic maneuverability simulation of the ship model is established.Secondly,build the measurement system model,the simulation of the measurement system model is established.The measurement data is pretreated based on the extended kalman filter.The possible fault types simulation can be introduced in the measurement system of dynamic positioning vessel.Thirdly,the method based on the neural network predictor theory is used in the fault diagnosis of the measurement system.This method is stable and reliable,effectively improve the fault recognition rate and detection efficiency rate of ship compasses.Finally,the support vector machine(SVM)method is used to detect sensor fault and make the appropriate diagnosis,through the binary tree method makes complicated multi class classification problem is simplified into a few relatively simple binary classification problem,reduce the difficulty of design.Then,The method based on the neural network predictor and the support vector machine theory is used in the measurement system diagnosis.Simulation results show the effectiveness of the proposed method in solving the measurement system diagnosis problem.