Evaluation Method of Ship Equipment Applicability Based on Agent Attention Transformer

This paper proposes a maritime shipborne equipment applicability assessment method based on agent attention Transformer to address the limitations of traditional assessment methods. As the application of artificial intelligence in the military field continues to expand, the assessment of new equipment applicability has become particularly crucial. This paper constructs a meshed index system that follows principles of comprehensiveness, hierarchy, and objectivity, dividing equipment applicability into four dimensions: reliability, maintainability, supportability, and availability, and further subdividing them into specific third-level indicators. Based on the Agent attention Transformer, the agent attention mechanism optimizes feature vectors and combines them with the triplet center-cross entropy loss function, effectively enhancing classification effects and recognition rates. Simulation experiments have verified the effectiveness of the proposed method, achieving an accuracy rate as high as 99.54%, and further confirming the model's superior performance in ablation experiments. The application of this method provides important data support for maritime cluster collaborative tasks and has significant military application value.