Research Of Key Technology For Multi-UUV System Working

With the continuous development of UUV technology,and the ocean development facing more and more complex issues,multi-UUV system working is becoming one of the hotspots in UUV research field.In its modeling techniques,there are three problems: only one single fitness function in traditional centralized task assignment having difficulty in meeting the multiple evaluation criteria,the traditional effectiveness evaluation methods having difficulty in transforming qualitative and quantitative indexes into a unified evaluation scale,and the description of intelligent UUV modeling technology imperfect.Therefore in this paper,a multi-UUV system task assignment strategy based on particle swarm optimization,a multi-UUV system effectiveness evaluation method based on cloud model and a UUV modeling technique based on social relationship and Agent are proposed.These effectively achieve the task assignment,effectiveness evaluation and intelligent description of multiple UUV system working.The main research findings and innovations are as follows:(1)In the multi-UUV system task assignment issue,with the system scale expanding and tasks increasing,the complexity of centralized task assignment has increased,and it requires a lot of computational and time costs;at the same time,the multi-UUV system task assignment needs to follow several evaluation criteria,but the traditional particle swarm optimization has only one single fitness function,can not meeting the requirements.To solve these problems,in this paper,the centralized hierarchy in multi-UUV system is designed and multi-UUV system and task sets are devided in hierarchy,the subsystem scale and assignment complexity reduced with the hierarchical reduction;and a multi-UUV system task assignment strategy based on multi-fitness-function particle swarm optimization,expanding the only one fitness function to multiple fitness functions,based on which,the inertia weight,learning factors and the particle update formulas in the traditional particle swarm optimization have been improved.The instance test demonstrates that this strategy can meet the multiple evaluation criteria in the task assignment,and take advantage of the high quality particles to jump out of local optimal solutions and achieve the global optimal solution.(2)In the multi-UUV system working effectiveness evaluation issue,the evaluation index system is the basis.Multi-UUV system collaborative operation taken as an example,for the shortage of the exsiting evaluation index system not including enough indexes,considering all the factors impacting multi-UUV system working effectiveness,the effectiveness evaluation can be devided into two parts: UUV individual and multi-UUV collaborative effectiveness evaluation,and the evaluation index systems for them is established.These two systems both include the qualitative indexes described with natural language and the quantitative indexes with precise values,but these two kinds of index cannot be evaluated and analyzed together.To solve the problem,the effectiveness evaluation method based on cloud model is improved.In this improved method,transition functions are proposed to deal with quantitative indexes,which transforms qualitative and quantitative indexes into cloud models in a unified evaluation scale,and in order to highlight the fuzziness,the concept of weight deviation is expanded to the concpt of weight deviation interval to represent the system effectiveness.An evaluation example for multi-UUV system collaborative operation effectiveness demonstrates the improved method is feasible and effective.(3)Multi-UUV system consists of several intelligent UUVs,and the UUV intelligent modeling research is the basis of multi-UUV system intelligence research.During the process of multi-UUV system executes the working tasks,there are several interacitons such as cooperation,consultation,competition,confrontation and so on between UUV individuals.In the Agent research based on social relationship,formal methods has difficulty in logic proofs and realization and there is a contradiction between the top-down design idea of informal methods and the Agent research idea.To solve this problem,the formalized definitions of individuals and society are given,and based on the analysis of the mapping between individual-society and Agent-MAS,the Agent model based on social relationship is given with the bottom-up method,including the concepts of Agent roles,role interactions,Agent organizations and so on.For the lack of the traditional BDI model expressing social characters,the expanded BDI model based on social relationship and the Petri-net description of this expanded model is given.The multi-UUV system anti-submarine warfare taken as an example,the funcitons and structures of UUVAgent based on social relationship is designed and the test example demonstrates that the design method effectively realize the UUV intelligent modeling.