基于改进实例推理的话务故障诊断专家系统

传统的K最近邻算法(KNN)算法可以解决话务分析专家系统中的求解问题,但KNN算法的不足在于K值的确定与执行效率,因此改进K值选取与加权方法,对提高算法运行效率与准确性具有重要意义.本文提出了一种改进K值选取方法及依托频率的权重计算方法,用于实例检索,并采用改进后的实例推理,构建了话务故障专家系统.实验结果表明,改进算法在实例匹配准确性与执行效度上,均优于传统方法.

Traffic Fault Diagnosis Expert System Based on the Improvement of Case-based Reasoning

Traditional K-Neavest Neighbor （KNN） algorithm resolves the problem of solving in traffic analysis expert system, but the disadvantage of KNN is the determination of the K value and the implementation of efficiency. The improvement of the selec- tion and weight method of K value are significant in enhancing the running efficiency and accuracy of the algorithm. A selection method for improving K value and weight algorithm method relying on the frequency are presented, and applied to case retrieval. The expert system of traffic fault diagnosis is built by improving case-based reasoning. The experimental results show that the improved al- gorithm are better than that of traditional methods in the case matching accuracy and the validity of execution.