Research On Emergency Decision Based On Case-Based Reasoning And Logic-Based Reasoning

In recent years, various kinds of emergent crises arise frequently, which have caused tremendous losses. Meanwhile, current crisis emergency decision infrastructure is not effective enough, emergency decision model and decision support system still need further improvement. Considering this situation, the researches on effective crisis management and efficient decision response realized by emergency decision model with computer aid system have become hot subjects for all related circles.Based on the researches of emergent crisis management and emergency decision, this thesis compares and analyzes prevailing emergency decision models and their advantages and disadvantages. A new emergency decision model based on case-based reasoning and logic-based reasoning is then proposed to make full use of the advantages of case-based reasoning including easy knowledge acquirement, fast reasoning speed and good learning ability while retaining the advantages of logic-based reasoning including clear representation structure, strong explanation ability and knowledge discovering ability. This model can satisfy the special requirements of emergency decision and improve effect and efficiency of decision process. Grounded on the establishment of the emergent crisis decision model, this thesis elaborates the realization of key steps of emergency decision and corresponding algorithms in detail. An object-oriented knowledge representation structure with multiple layers is introduced firstly. Then, supplementary rules and an integrated attribute similarity calculation method are used together to improve the case retrieval algorithm, which helps to establish a parallel knowledge search mechanism with the logic reasoning algorithm realized by Prolog. After that, transformational adaption strategy and group decision strategy supported by computer system are combined to design an effective case adaption process. At last, a three-layer value evaluation algorithm is introduced to realize system self-learning and a rule coverage method is used to deal with rule redundancy.With the proposed emergency decision model and realization methods, a sample flood disaster emergency decision process is presented to introduce the application of the emergency decision model and related techniques in practice. At the end of the thesis, key conclusions and further research fields are also discussed.