Research Methods Of Filtering For UUV's Forward Looking Sonar Based On Bio-inspired Algorithms

In the process of performing tasks in the shallow water, UUV makes online rolling route planning and collision avoidance response based on the local environment information probed by the forward looking sonar (FLS). Because of the complexity of the physical environment in shallow water and the sound fields, the information probed by FLS was easy to be affected by geology of shallow water, hydrology, environment noise, multi-path effects near the surface of water, etc. Which made that there has much noise and counterfeit objects among the probed information, influenced UUV's online rolling route planning, and even threated the security of UUV. To "eliminate the false and retain the true",and extract the credible obstacle information from the original FLS probed information, this paper combined with UUV motion characteristics and handling characteristics, introduced bio-inspired algorithms to study the FLS information filtering algorithms. Mainly studies were as follows:First, established the sight model of UUV's FLS, analyzed the data characteristics of FLS under the interference of shallow water environment. On the basis of FLS operating parameters,we established the sight model of UUV's FLS, discussed the representation methods of data structure for the information of object in the sight of FLS; analyzed the influence from FLS operating parameters, shallow water environments and multi-path effects to the probed information, established the credible model of probed information by FLS.Then, we drew up a range criterion for the FLS information, introduced humanoid observation inspired data processing algorithms for UUV's FLS data, and realized noise processing combined with wavelet transform. On the basis of the FLS sight model and the analysis of UUV's obstacle avoidance response, it formulated a range criterion for FLS information; by discussing the influence from UUV's route planning to the process of FLS probing, a humanoid observation inspired strategy was introduced for FLS data processing,and incorporation algorithm of a sliding window and fuzzy clustering analysis was proposed to realize the identification of object data.After that, we discussed the environment modeling for UUV by grid, researched the attribute detection and motion prediction of the obstacle in the sight of UUV, and proposed an improved D-S updating algorithm for grid map. Aiming at the characteristics of FLS information and shallow water environment, UUV's local grid mapping based on FLS information was discussed; according to the relevance between FLS information and grid map,the attribute detection and motion prediction of the obstacle were realized; during the updating process of grid map by D-S fusion algorithm, the credible model of FLS probing was introduced to improve the credit assignment,and a threshold decision method was proposed to achieve the updating of grid map.Finally, according to the motion characteristics and handling characteristics of UUV, we designed the strategy for UUV route planning and the methods for avoiding obstacle,proposed a route planning and avoidance method on the basis of an improved biofilm neural network algorithm. On the basis of the motion property of obstacle, we did the corresponding collision avoidance methods research for UUV; when using the self-adaption characteristics of the biofilm neural network to plan path, the constraints of UUV motion was considered in,and UUV's collision avoidance was achieved by combining the proposed avoidance algorithm on the basis of collision avoidance detection mechanism. By designing the simulation of route planning and collision avoidance for static obstacles and dynamic obstacles, the algorithm proposed was verified effectively...