The Research And Application Of Multi-sensor Real-time Information Processing Algorithm In Sea

It is of great significance to build an oceanographic information awareness network in maintaining the security of China's territorial waters,monitoring the order of marine traffic,developing the marine economy and creating a leading international marine information system.With the development of information technology and the construction of the blue sea information network system,in order to get better perception of the target,the measurement information of a single sensor cannot meet the requirements.So as to get the global target motion state,the information that provided by a wide variety of sensors should be processed globally.Based on the data fusion processing system of "blue sea information network",the problem of data comprehensive processing in marine multi-sensor networking system is investigated.In this thesis,three sensors i.e.radar,Automatic Identification System(AIS)and Automatic Dependent Surveillance-Broadcast(ADS-B)are introduced to monitor the targets,such as aircraft and ships.In this thesis,the working principle,mode,distinguishing feature and message structure of the AIS,ADS-B and radar are introduced firstly.And based on the advantages and disadvantages of target detection of three sensors,the complementarity and feasibility of the data fusion processing system is analyzed.For multi-sensor data processing,this thesis first introduces the preprocessing of the data,which includes three parts: error data elimination,time synchronization and coordinate transformation.Aiming at the multi-radar network track association model,a track association algorithm based on graph and B-type gray association is proposed.In order to reduce the association times and improve the track association efficiency,the topological legends of multi-radar network and double threshold detection is introduced,and therefore the track association is only carried out between the adjacent tracks.Compared with the traditional nearest neighbor method and the weighting method,the accuracy of track correlation is improved obviously.And compared with the fuzzy association algorithm,the correct correlation rate of the track is increased by 2.47%.In general,the validity of the algorithm is verified.For the multi-sensor track association at the sea,this thesis proposes a three level track association strategy,that is,the local tracks are directly related to the system tracks,the local tracks are indirectly related to the system tracks and the tracks association between the local tracks.And the basic information of each sensor's ID and local track batch number are incorporated into the track association algorithm,which reduces the unnecessary association calculation and the time consuming.In addition,the three level track association strategy that proposed in this thesis has been applied to the simulation demonstration system of this subject.At the end of this thesis,the processing of multi-sensor track fusion is introduced.For high precision AIS and ADS-B sensors,these local measured tracks are used as the system fusion tracks,and three fusion algorithms are introduced for the track fusion processing between low precision radar,including directed mean method,weighted mean method and dynamic weight allocation method.Finally,the simulation test shows that the dynamic weight allocation method has better fusion precision compared with the direct mean method and the weighted mean method in the complex sea detection environment.