武器效能衍生推理及评估模型自学习的研究实现

Derivative Reasoning of Weapon Effectiveness and Self-learning of Evaluation Model

针对海洋环境下武器作战效能的评估问题,在样本数据极度匮乏的情况下,提出了一种武器效能衍生推理机制。利用与待评估武器具有较高相似度的一种或多种其它武器型号的样本数据,训练通用评估模型得到针对该武器的专用评估模型。为提高评估模型的可信度,专用评估模型增加了自学习功能。随着武器样本数据更新变化到一定阈值,对模型进行重新训练以完成专用模型的自学习修正。最后在某作战平台效能评估系统上进行实例验证并完成模型自学习功能设计。事实表明衍生推理能提供较高参考价值,并且通过模型自学习,效能评估的可信度和准确性有很大提高。

Aiming at the operationaI effectiveness evaIuation probIem of weapon equipment under the marine environment,this pa-per proposes a derivative reasoning mechanism of weapon operationaI effectiveness in the condition of extreme scarcity in the sampIe data.Using sampIe data of one or more other types of weapons having a high simiIarity with the weapon to be evaIuated,a common evaIuation modeI can be trained to get speciaI evaIuation modeI for the weapon.To improve the reIia-biIity of evaIuation modeI,the function of seIf-Iearning is added to speciaI evaIuation modeI.With sampIes changing to a cer-tain threshoId,the seIf-Iearning revision of speciaI evaIuation modeI can be compIeted by modeI retraining.