计算机与通信技术ComputerandCommunicationTechnology《自动化技术与应用》2023年第42卷第6期TechniquesofAutomation&Applications基于优化决策树算法的变电站故障诊断系统研究*李杰,孙鹤林,雷一鸣,田晓雷,蔡正梓（国网北京市电力公司，北京100000）摘要:为有效提升变电站运行期间故障自动化诊断效率和准确率，采用阈值近邻迭代法对样本数量进行优化，同时引入平衡系数对测试属性选择进行优化，然后再采用优化二分离散算法对连续属性离散性进行优化，构建基于优化决策树算法的变电站故障诊断系统。通过优化决策树、ID3以及C4.5三种算法的应用效果对比分析，得出基于优化决策树算法分类准确性更高、决策树构建速度更快、决策树规模更小这一结论，能够显著提升监控系统在变电站运行期间的应用效果，对于变电站长期运行稳定与安全具有重要意义。关键词:优化决策树算法；故障诊断系统；阈值近邻迭代法；平衡系数；优化二分离散算法中图分类号：TP277文献标识码:A文章编号:1003-7241(2023)06-0112-05ResearchonSubstationFaultDiagnosisSystemBasedonOptimizedDecisionTreeAlgorithmLIJie,SUNHe-lin,LEIYi-ming,TIANXiao-lei,CAIZheng-zi(StateGridBeijingElectricPowerCompany,Beijing100000China)Abstract:Inordertoeffectivelyimprovetheefficiencyandaccuracyofautomaticfaultdiagnosisduringsubstationoperation,thethresholdnearestneighboriterationmethodisusedtooptimizethenumberofsamples,andthebalancecoefficientisintroducedtooptimizetheselectionoftestattributes.Thentheoptimaltwoseparationalgorithmisusedtooptimizethediscretenessofcontinuousattri-butes,andthesubstationfaultdiagnosissystembasedonoptimaldecisiontreealgorithmisconstructed.Throughthecomparativeanalysisoftheapplicationeffectofoptimizeddecisiontreealgorithm,ID3algorithmandC4.5algorithm,itisconcludedthattheclassificationaccuracyofoptimizeddecisiontreealgorithmishigher,theconstructionspeedofdecisiontreeisfaster,andthescaleofdecisiontreeissmaller,whichcansignificantlyimprovetheapplicationeffectofmonitoringsystemintheoperationperi-odofsubstation,andhasimportantsignificanceforthelong-termstabilityandsafetyofsubstationoperation.Keywords:optimaldecisiontreealgorithm;faultdiagnosissystem;thresholdnearestneighborite...
