摘要
针对由混合配置前件导致的大量耦合的、相互关联的复杂规则,提出了LC神经网络和RULEX算法联合实施下基于历史数据的个性化产品服务配置规则获取方法,包括基于LC的规则构造和知识发现,基于RULEX的规则抽取,网络模块的划分与提炼,以及配置规则适应度评价。最终通过楼宇控制产品服务说明了该方法具有较好的网络训练效率和规则复杂度。
To deal with a large number of coupled,connected complex rules caused by mixed antecedents,acquisition methods for personalized product service configuration rules based on LC neural network and RULEX algorithm is proposed in this paper. First, rule knowledge is discovered by LC from historical data, then extracted by RULEX. Network module partition and analysis of rules fitness are also presented. Finally, the proposed method is validated by services of building control product.
出处
《工业工程与管理》
CSSCI
北大核心
2012年第3期66-73,共8页
Industrial Engineering and Management
基金
国家自然科学基金重点项目(70932004/G0209)
