针对故障诊断过程中冗余征兆问题,提出了一种启发式遗传约简算法;该方法将核加入到遗传算法的初始种群中来提高算法的性能,引入一种信息论角度定义的属性重要度度量作为启发信息,构造修正算子引入启发信息,使得被选择的属性子集的分类能力不变,从而保证群体进化收敛于最小约简;最后以某汽车发动机故障诊断决策表为例,结果表明,该算法可以有效地对故障征兆进行约简,能够提取出最能反映故障的特征,从而为粗糙集在故障诊断中的深入应用打下了基础。
For the problem of redundant symptoms in fault diagnosis, an heuristic genetic algorithm for minimal attributes reduction is presented. The core is brought in initial population in GA to improve its performance. The significance of attributes defined in the information theory is imported into this algorithm as heuristic information. To ensure the ability of classification of the attributes set, a new modification operator is constructed . Finally, the experimental results show that the algorithm can effectively obtain the minimal reduction of the fault symptoms of automotive engine .