| With the technology change rapidly people scramble for marine resources becomes more and more fierce, ship dynamic positioning technology provides an effective means to develop for marine resources, but how to improve DP (dynamic positioning) ship operation safety, the positioning accuracy and reliability has been the focus of this technology. In the sea, due to the complexity and variety of uncertain factors, the effect of depending on a single sensor to realize localization is always barely satisfactory, therefore the multi sensors data fusion has become a research focus. This study is targeted at the dynamic positioning vessel,according to the statistical characteristics of sensor measurement noise is not accurate,a fuzzy adaptive single sensor filtering algorithm was proposed to get the vessel's position of state estimation.Then using multilevel hierarchical structure, comprehensive utilization of the information of each sensor, to get more accurate estimation of ship motion state, this provides a reliable guarantee for realizing the accurate location of the ship.In this paper, firstly, the dynamic system model is built, including the ship motion model and the sensor observation model. Among them, the ship motion model including high, low frequency motion model of ship, and the model of the ocean environment load, sea wind,wave and current. The sensor observation model is mainly for the ship position measurement system of differential GPS, Hydroacoustic Position Reference and Taut Wire system. The simulation results show that the model is correct, and this chapter provides necessary conditions for the follow-up studies.The measurement noise's statistical characteristics of dynamic positioning sensor measurement system varied with the actual working environment, according to the statistical properties of measurement noise possibly inaccuracy, this paper proposes an adaptive kalman filtering algorithm based on fuzzy logic. Firstly, the basic principle of Kalman filter to liner filtering method is introduced, then proposed fuzzy adaptive kalman filter, by comparing whether the ideal residual and actual residual is consistent, and the degree of consistency,using fuzzy inference system to get the adjusted coefficient of measurement noise covariance matrix and then adjust the matrix, so that the filtering precision is proved, the divergence of filtering is suppressed as well. Due to the dynamic positioning system is nonlinear, this method is extend to cubature kalman filter of nonlinear filter ,at the same time, in order to overcome the computational divergence, square root cubature kalman filter was proposed.Combined it with fuzzy inference system, at last the square root cubature kalman filter based on fuzzy adaptive filter was achieved, the measurement noise covariance matrix is adjusted online and the accuracy of the estimation of ship motion state is improved.To make full advantage of the measurement information provided by sensor subsystem,improve the position precision of state estimation, scientific and rational allocation of multi-sensor resources is needed. Federal filter as an important decentralized filter, its application is more, multilevel hierarchical fusion architecture can be seen as an expansion of the federal filter structure, the fusion accuracy is improved. In this paper, the federated filtering algorithm and multilevel hierarchical data fusion algorithm is studied, and compared the accuracy of two information fusion method.Finally, the overall simulation system of dynamical positioning system is built, the study of multi sensor data fusion algorithm based on fuzzy adaptive filter is simulated. From the results, it can be shown that the algorithm proposed in this paper is feasible, improves the estimation accuracy of ship position, ensures the safety, reliability and overall positioning accuracy of DP system. |
PDF Full Text Request