摘要
为解决因训练数据量匮乏导致深度学习大模型泛化性能差的问题,提出了一种引入装备基础理论知识作为辅助驱动的装备效能智能预测方法。首先,概述了通用装备基础模型构建范式,并阐述了将理论知识引入深度学习模型的训练方法;然后,针对效能预测涉及的多维度时间序列预测问题,提出了GA-CNN-LSTM混合预测模型;最后,以典型预警雷达装备的作战仿真效能评估为例,对混合预测模型进行了试验验证。试验结果表明,混合预测模型在R平方分数性能指标上比原始模型提高了约2.9%,模型有效且实用。
In order to solve the problem of poor generalization performance of deep learning large model due to lack of training data,an intelligent prediction method of equipment efficiency is proposed by introducing basic equipment theory as suxiliary driver.Firstly,the general equipment basic model construction paradigm is summarized,and the training method of introducing theoretical knowledge into deep learing model is expounded.Then,aimed at the multi-dimensional time series prediction problem involved in efficiency prediction,a GA-CNN-LSTM hybrid prediction model is proposed.Finally,taking the operational simulation effectiveness evaluation of typical early warning radar equimment as an example,the hybrid prediction model is tested and verified.Experimental results show that the performance index of the hybrid prediction model improves the R2 score performance metric by approximately 2.9% compared to the original model,and the model is effective and practical.
作者
周家豪
杨学康
郭伟然
黄翔
张捷
ZHOU Jiahao;YANG Xuekang;GUO Weiran;HUANG Xiang;ZHANG Jie(School of Automation,Nanjing University of Science and Technology,Nanjing 210094,China;Science and Technology on Information Systems Engineering Laboratory,Nanjing 210023,China)
出处
《指挥信息系统与技术》
2023年第6期39-47,共9页
Command Information System and Technology
关键词
理论辅助
数据驱动
效能预测
theory-assisted
data-driven
efficiency prediction
作者简介
周家豪,男(1999-),硕士研究生;杨学康,男(1998-),硕士研究生;郭伟然,男(1989-),博士,工程师;黄翔,男(1995-),助理工程师;张捷,男(1979-),研究员。
