Research On AUV Intelligence Evaluation Index System And Evaluation Method

With the development of artificial intelligence,navigation technology,and control theory,AUV has gradually developed from the traditional working mode to intelligent and unmanned.Intelligent AUV has become an important tool for marine development in the new century and is widely used in military,Civilian and other fields.The intelligence of the AUV directly affects whether the AUV can complete its tasks in harsh environments and whether it can ensure its own safety.Therefore,by evaluating the intelligence of the system,it can provide a reference for the intelligent development and design optimization of AUV.As there is no comprehensive evaluation of AUV intelligence,and there is no clear definition of AUV intelligence,the establishment of a reasonable AUV intelligence evaluation index system is the basis for obtaining reasonable evaluation results.The multi-expert weight method can reduce the subjectivity of subjective weights,and the method of solving multi-expert weight combination coefficients directly affects the rationality of the evaluation results.Therefore,it is of great practical significance to carry out research on the AUV intelligence evaluation index system and evaluation methods.The topic focuses on the needs of AUV intelligent evaluation,starting with the three aspects of index system construction,multi-expert weight determination and evaluation methods to carry out related key content research,focusing on the following research:(1)Starting from the existing problems of AUV intelligence evaluation,design problem solving ideas.Based on the analysis of AUV’s task development and the development of unmanned systems’ intelligence,the definition of AUV’s intelligence is given.The AUV’s intelligence is analyzed from three aspects:motion control,target recognition and path planning,and a preliminary indicator system is given.,And on this basis,the content of the index system is tested and the final AUV intelligent index system is obtained,which provides support for the follow-up evaluation method research.(2)In view of the subjectivity of the multi-expert weight method and its dependence on non-dimensional results,after analyzing the current main weight determination methods,information entropy and consistency test theories are introduced respectively,and a multiexpert weight single-objective optimization model is comprehensively established.Finally,in order to solve the single-objective optimization model,the DE algorithm is introduced,and through simulation experiments,the combination coefficients of the weights of multiple experts are solved.(3)Aiming at the problem of slow convergence speed of DE algorithm and easy to fall into local optimum.First,the concept of counterpoint learning is introduced,combined with the DE algorithm,and the process of the ODE algorithm is explained,which speeds up the convergence of the algorithm.Introduce an elitist idea,further optimize the ODE algorithm,and propose an OBL-EMSDE algorithm.Finally,through simulation verification,it is proved that the algorithm has superiority in the optimization ability and speed of unimodal function and multimodal function.(4)Aiming at the difficult selection of synthetic models in the evaluation process.Because the AUV intelligent system is a multi-level and multi-factor system,the fuzzy comprehensive evaluation method is introduced to establish a mathematical model of the fuzzy comprehensive evaluation method,so that the evaluation results can fully reflect the impact of the evaluation indicators.As the AUV intelligence evaluation process is a process of solving unknown systems through known information,the gray system method is introduced to establish an evaluation model for graying weight and fuzzy mathematics,and the evaluation results of the two methods are verified by simulation to be consistent and reasonable.