改进的舰船维修费案例调整模型

摘要

In order to enhance the adaptation effect of ship maintenance cost case, an improved ship maintenance cost case adaptation model is proposed according to the modeling process of producing adaptation knowledge, reducing adaptation knowledge and reasoning adaptation knowledge. In this model, the ship maintenance cost case adaptation knowledge is produced by the direct similarity sets and indirect similarity sets of target cases based on the Policastro's format; towards the problems that correlation and redundancy exist in the above adaptation knowledge, or the knowledge is too large to deal with, an improved rule reduction method is put forward which can both reduce the above knowledge effectively and measure their importance accurately, in addition, the algorithm for rule reduction is given; in order to improve the performance of knowledge reasoning, least square support vector machine (LSSVM) which has strong generalization power and perfect learning power is used to establish the ship maintenance cost case adaptation knowledge reasoning model, and the parameters of model are optimized according to genetic algorithm (GA) and cross-validation method. Example analysis reveals that, compared with other case adaptation models, the adaptation accuracy of the proposed model is far better, so the proposed model is scientific and valid.%为提高舰船维修费案例的调整效果,遵循着调整知识产生、调整知识约简和调整知识推理的建模过程,提出了一种改进的舰船维修费案例调整模型.在该模型中,采用Policastro的重构方式,利用目标案例的直接和间接相似案例集,产生舰船维修费案例的调整知识；针对调整知识中可能存在冗余、相关性、知识量过于庞大不便于推理等问题,提出了一种既能有效约简调整知识、又能准确衡量调整知识重要度的改进规则约简方法,并给出了具体的实现算法；为提高知识推理的效率,采用具有强泛化能力和出色学习能力的最小二乘支持向量机(least square support vector machine,LSSVM),构建了舰船维修费案例调整知识推理模型,并基于遗传算法(genetic algorithm,GA)和交叉验证方法,对推理模型的有关参数进行了优化.实例结果表明,与其他案例调整模型相比,所提模型在调整精度上要远远好于其他模型,所提模型是科学、有效的.