An MMCW-FCE Method for Evaluating AUV Intelligence on the Algorithm Level

Since an intensely subjective expert weights and the divergence of opinions among experts exist in the evaluation, this paper proposes a modified multi-expert combined weight (MMCW)-fuzzy comprehensive evaluation (FCE) method. This algorithm preliminarily screened the expert-weight set based on the cluster analysis. After designing the objective function according to the information entropy and relative entropy theory, the multi-expert combined-weight single-objective optimization model was constructed via the weighting coefficient. However, the solution happens to poor convergence in the traditional intelligent algorithm, easily falling into the local optimum. Thus, a differential brain storm optimization (DBSO) algorithm was exploited to figure the established optimal model out for the combined coefficient, and the result was obtained via the hierarchy fuzzy comprehensive evaluation. As taken the AUV as an example, the three-layer evaluation index system was established from the motion control, target recognition, and path planning to assess the AUV intelligence, and the simulations were conducted to validate the proposed algorithm. Conclusively, the MMCW-FCE can quantitively evaluate the intelligence of diverse AUV systems on the algorithm level.