Research On Shipboard Positioning Technologies Of Location Fingerprint Method

As the location-based service has found an increasingly wide and extensive utilization in different fields,there is an increasing demand for indoor location-based services in recent years.Being an important water transportation tool,ships are becoming increasingly large in volume and complex in structure,characterized by their multi-deck and large load capacity.However,it is rather difficult to apply common indoor positioning technologies in the cabin due to its complicated environment and different sailing factors.Therefore,how to solve the problem concerning internal information awareness and to provide theoretical and technical supports for the indoor location-based service is of practical significance and served as an important direction in current field of indoor location.This article mainly study on indoor localization technologies in the ship environment.Firstly,we introduce the traditional indoor positioning methods.Comparing the advantages and disadvantages of distance measuring based positioning technologies and wireless signal strength based positioning method.Mainly studies the indoor positioning technology based on fingerprint positioning.For analyzing the characteristic of the Marine environment of the wireless signal,we select the Yangtze No.2 ship to do researches.Finally,by using ZigBee sensor network platform,we conduct a series of wireless signal experiments on the ship.Marine environment were analyzed by using statistical distribution characteristics of the wireless signal of experimental.The navigation in shipboard were studied under time-varying characteristic of wireless signal,summarizes the environment's influence on the position fingerprint positioning method in shipboard.Secondly,in order to reduce the influence from marine environment to the position fingerprints.This paper proposes a space-time location fingerprint feature extraction method,through space difference build stable signal database,then use the principal component analysis and linear discriminant analysis for dimension reduction and the sample clustering database.Under the premise that keep enough position characteristic information,and filter the irrelevant noise signal and the redundant information from the formal data.These features reduce the amount of calculation algorithms,improving the positioning precision of the position fingerprint.With time-series analysis method at the same time,the fingerprint features increase the stability of the fingerprint characteristics.After the characteristics of all kinds of fingerprint visualization and algorithm testing,we verify the stability and accuracy of the fingerprint characteristic.Finally,we study the statistical learning methods,and put forward the position fingerprint positioning method based on support vector machine(SVM),which can quickly identify the fingerprint with the fingerprint search and matching.At the same time,we use kernel function to solve the problem of the nonlinear classification of the position fingerprint.By using cross validation and grid-search,we can find the optimization of parameters which can obtain the optimal parameters of model training.After input the training sets into support vector machine(SVM)and modeling,we can get a decision function for the online fingerprints indentifying.The testing experiments are implemented in shipboard.In this paper,we use artificial neural network and the original data to train and test the positioning accuracy,which verify the less amount of calculation of our algorithm.The methods proposed in this paper can satisfy the precision of positioning shipboard which demanded by the location-based service applications.