摘要
运用多项数据分析及推理技术提高物资需求预测速度及可靠性.首先利用历史案例信息求救援案例指标权重;之后建立模糊聚类(FCM)及案例检索相结合的算法,案例检索采用CBR-GRA双重检索技术,在得到相似度向量与灰色关联度向量之后,再次应用灰色关联分析求取案例相似-关联度向量,从而保证可靠案例检索;最后建立救援物质需求模型.经实例验证可知:案例聚类实现数据初步筛选,提升了检索速度,2种检索方法融合,提升了检索可靠性.
Multi-data analysis and reasoning techniques were adopted to improve the forecasting speed and reliability of emergency resources demand. Firstly, based on the historical case information, the rescue case index weights were calculated. Then an algorithm combining fuzzy C-means clustering with case retrieval was established to increase the efficiency of case retrieval, which was performed by CBR （casebased reason） similarity and GRA （grey relational analysis） correlation. After the CBR similarity vector and GRA correlation vector were obtained, the grey relational analysis was used to calculate the similarity-correlation vector so as to ensure that similar cases are retrieved efficiently. Finally, a resources demand model was built up. The results confirmed that case clustering to achieve preliminary data filtering can enhance retrieval speed and combining two retrieval methods can improve the reliability of retrieval.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2016年第5期756-760,共5页
Journal of Northeastern University(Natural Science)
基金
"十二五"国家科技支撑计划项目(2012BAK13B01)
关键词
应急救援
需求预测
案例推理
灰色关联分析
模糊C均值聚类
主客观综合权重
emergency rescue
demand forecast
casebased reason （CBR）
grey relational analysis （GRA）
fuzzy C-means clustering
subjective and objective comprehensive weight
