基于决策树的轨道电路故障诊断知识表示方法研究

针对ZPW-2000A无绝缘轨道电路故障逻辑机理模糊的问题,本文采用了基于决策树的轨道电路专家系统知识表示方法.该方法首先将轨道电路故障影响较大的特征向量样本建立故障决策表,然后运用最小信息熵算法对属性值离散化,根据决策树算法快速学习及分类的特点对离散后的数据样本进行训练学习,生成故障决策树后进行知识规则的获取,在专家系统的知识库中以产生规则存储.通过对ZPW-2000A无绝缘轨道电路的实例分析验证了该方法在轨道电路专家系统知识表示与获取中的有效性和实用性.

The Research of Fault Diagnosis Knowledge Representation of Track Circuit Based on Decision Tree Method

For the problem of ZPW-2000A jointless track circuit fault fuzzy logic mechanism,this paper adopts the track circuit expert system knowledge representation method based on decision tree. This method first samples of a greater influence on the track circuit fault feature vector to build up the fault decision table,then use the minimum information entropy algorithm to discretize attribute value,according to the characteristics of fast learning and classification of the decision tree algorithm,it trains and learns discrete data samples,obtains knowledge rules after generating fault decision tree,then the rules are stored in the knowledge base of expert system . Through the instance analysis of ZPW-2000A jointless track circuit,it verifies the method is validity and practicability in the expert system knowledge representation and acquisition of track circuit.