基于时间序列的柴油发电机组故障录波数据增量式挖掘算法

Incremental Mining Algorithm for Fault Recorded Data of Diesel Generator Based on Time Series

空压机房和临时备用发电机房通常选用较为吸音的隔音材料设防,难以获取故障录波数据,挖掘精度较低,为此提出了基于时间序列的柴油发电机组故障录波数据增量式挖掘算法。通过柴油发电机组故障录波数据的预处理,结合等长处理定义故障录波数据的时间序列匹配度,利用故障录波数据的信息状态函数,提取故障录波数据的时间序列,根据柴油发电机组故障的具体情况,获取柴油发电机组故障的位置,结合故障录波数据的清洗和无量纲化处理,构建柴油发电机组故障录波数据回归方程,通过定义故障录波数据的聚类中心,建立故障录波数据的权重映射矩阵,根据提取出的故障录波数据的特征,挖掘出柴油发电机组故障录波数据。实验结果表明,该算法具有更高的加速比和挖掘精度,当融合系数为0.8时,故障录波数据的挖掘输出相对稳定,可达到90%以上。该方法能够挖掘出柴油发电机组的故障录波数据。

The air compressor room and the temporary standby generator room are usually fortified with sound-absorbing sound insulation materials,so it is difficult to obtain fault recording data and the mining accuracy is low.Therefore,an incremental mining algorithm for fault recording data of diesel generator sets based on time series is proposed.Through the preprocessing of the fault recording data of the diesel generator set,the time series matching degree of the fault recording data is defined by the isometric processing,and the time series of the fault recording data is extracted by using the information status function of the fault recording data.Combined with the cleaning and dimensionless processing of fault recording data,the fault recording data regression equation of diesel generator set is constructed.By defining the clustering center of the fault recording data,the weight mapping matrix of the fault recording data is established.According to the characteristics of the extracted fault recording data,the fault recording data of diesel generator sets are excavated.Experimental results show that the algorithm has higher speedup ratio and mining accuracy,and when the fusion coefficient is 0.8,the relative stability of the mining output of fault recording data can reach more than 90%.This method can mine the fault recording data of diesel generator sets.