Research On Demand Forecasting Of Emergency Supplies Based On Fuzzy Rough Set And CBR

In order to improve the efficiency of emergency relief work and to minimize the number of casualties and property losses,the primary task is to provide relief supplies to the disaster area.However,demand for emergency supplies can not be accurately obtained because of sudden,non-routine and uncertain unconventional features of unexpected events,causing supply imbalances,and making limited emergency supplies can not be used efficiently.It also brings difficult for demand forecasting of emergency supplies that emergency-related information is incomplete or can not be obtained accurately.Fuzzy rough set can handle imprecise information,dealing with inconsistencies between condition attributes and decision attributes.Fuzzy rough set is used for attribute reduction and weight distribution,getting rid of the condition attributes that less affect decisions,while retaining the core attributes and assign weights to the core attributes.And examples of earthquake are used to demonstrate the feasibility and rationality of the method that can solve the problem of demand forecasting of emergency supplies well,in the case of incomplete informationIt also needs to store cases to achieve demand forecasting for emergency supplies after the irrelevant attributes are eliminated.Case storage is the basis of case-based reasoning which is used to achieve the demand forecasting for emergency supplies.And the basis of case storage is case representation and organization.Hybrid ontology and attribute are combined to achieve tiered case representation and XML format is used to store documents.The purpose of case storage is to retrieve the most similar source cases.It requires a certain search strategy for case retrieval,and similarity calculation is the core of the case retrieval.When calculating partial similarity of cases,different methods are used for numerical attributes and object properties.And then calculate global similarity through weighting the partial similarity.A final solution can be gained through adjusting the retrieved similar cases.Case adjustment methods of local case adjustment and global case adjustment are used to adjust the cases.Finally,system framework of demand forecasting for emergency supplies based on case-based reasoning and ontology is built,combined with rule reasoning to achieve case retrieval and case adjustment,achieving demand forecasting for emergency supplies.Based on the frame structure,combined with professional knowledge of the field of emergency supplies demand,domain ontology of emergency supplies demand can be built with Protege.Firstly,inference engine Racer is used to check the consistency of ontology.And then SWRL is used to establish one's own rules,implementing case-based reasoning and rule-based reasoning.Case-based reasoning software myCBR is used to implement case-based reasoning,embedding myCBR into Protege,achieving integration of ontology and case-based reasoning.Inquiring cases with myCBR,the inquired results will be returned by myCBR,providing a suitable demand proposal for emergency supplies.Finally,prototype system of demand forecasting for emergency supplies is built on the platform myeclips8.0.The system include five functions of landing system,adding cases,modifying cases,deleting cases and retrieving cases.