Muti-stage Fusion Method For Measure Data From Dynamic Positioning System

Along with the economic development,the scarcity of natural resources is always the problem that people highly focus.Due to land resources is limited.The ocean,which occupy vast areas of earth,has become the major source of life needed energy that should find and exploit.Ship dynamic positioning technology plays a key role in the protection of offshore exploit,it's also a hot research to improve the accuracy of dynamic positioning technology .Because of the accuracy requirement of dynamic positioning and the instability of complex maritime environment,we need to install a variety of sensors to get more measurement data.It lead to using a variety of sensor data fusion to estimate the state of the ship became the focus of research. In this paper, under the background of nonlinear ship and supported by the project called development and industrialization of Ocean engineering vessel dynamic positioning system.well study on two algorithms named Multi-level fusion method and weighted fusion method.When driving the multi-sensor data fusion,it will increase the number of dimensions and time of calculation if we use centralized algorithm.Otherwise,it can contaminate the entire set of data when sensor is malfunction. Therefore,there are a lot of studies related to hybrid algorithm currently.In this paper,first of all,according to the motion equations of ship sea level,a ship motion mathematical model will be built.Based on the principles of the reference position,we'll build tension cords, acoustic positioning, differential GPS (DGPS) and laser positioning system models.Furthermore, gyrocompass, motion reference unit and wind sensors related program module will be designed to estimate the condition of the ship.Secondly, the multi-level fusion method will be studied.Using the sub-filter that adopt unscented Kalman filter algorithm to carry the local optimal estimation of sensor measuring data.Besides,Using the primary filter that adopt optimally weighted algorithm to get the global optimal estimation by calculate local optimal estimation of sub-filter.In the process of the local optimal estimation step,this paper proposes an improved multi-level fusion method.Apply the thought of Euclidean distance and cluster analysis to reorder the sensor measurement data,then determine the order of integration.By letting the accurate data fusion priority and increasing the value of weight in main filter,multi-level fusion algorithm have some improvement in the accuracy than before. And the distance matrix between the sensor data which calculated by the Euclidean distance algorithm is different from distance determination in previous studies.lt reducing the error based on the characteristics of data itself,rather than selected parameters by human experience.Finally, completing the development position measurement system,including serial port driver module, sensors packet analysis module, data preprocessing module and the core algorithm module.The core algorithm module consist of improved multi-level fusion algorithm and weighted average algorithm.Both algorithms and data processing results will compared and analyzed.Furthermore,Comparison and analysis of multi-level fusion algorithm simulation also will be carry on to verify improved fusion algorithm of multi-stage projects can be realized and accuracy.