基于情景推理的震灾初期救援物资需求预测

Demand Forecasting of Relief Materials for Early Post-earthquake Rescue Based on Scenario Reasoning

震后救援物资响应是救援工作的重要环节,考虑到震后初期信息匮乏与案例推理预测结果精度低,采用情景推理对救援物资需求进行预测:通过对震灾历史案例库的分析,构建知识元表示模型对震灾情景进行表达。基于云模型启发式算法获取情景检索特征属性的权重,再运用改进的最近邻检索算法对案例间情景进行检索,最后利用派生重演法对检索出的最相似案例进行重用修正。以中国地震历史数据作为实验分析具体案例,结果表明本文方法对提高灾后应急物资预测精度具有良好的效果。

The paper pointed out that,in the event an earthquake,considering the lack of information and the low precision of case based reasoning prediction results in the early post-strike stage,the paper used scenario reasoning to predict the relief materials demand.By analyzing earthquake histories in case library,the paper constructed a knowledge element representation model to express the earthquake scenarios.Based on the heuristic algorithm of the cloud model,it obtained the weights of the scenario retrieval feature attributes,used the improved k-nearest neighbor(KNN)algorithm to retrieve the scenarios among cases,and used the derived recapitulation method to reuse and correct the most similar cases retrieved.The paper took the historical earthquake data in China as the specific cases for experiment,which showed that the proposed method has good effect on improving the prediction accuracy of post-earthquake emergency supplies.