装备故障的时空共现模式挖掘

Mining Spatiotemporal Co-occurrence Pattern in Equipment Failure

针对装备故障数据的时空特性,给出一种基于apriori算法的快速挖掘算法。从装备故障全局的角度出发,定义装备故障时空共现模式,模式中引入故障实例空间参与率、候选同位模式参与指数、故障类型时域参与度、故障类型时域参与指数等描述指标,并通过仿真分析比较快速挖掘算法与朴素挖掘算法的执行效率。仿真结果表明:当装备故障时空数据量较大且含较高噪声时,所提出的快速挖掘算法有更高的执行效率。

For spatiotemporal characteristics of the equipment failure data, a fast miner algorithm is proposed based on apriori algorithm. From a global perspective of equipment failure, equipment failure spatiotemporal co-occurrence pattern is defined. The pattern is described with several parameters, which are failure-instances spatial participation ratio（SPR）, candidate spatial co-location pattern participation index（SPI）, failure-types temporal participation degree（TPD） and failure-types temporal participation index（TPI）. By simulating, analyzing and comparing with execution efficiency between the fast miner algorithm and naive miner algorithm, we find that the fast miner algorithm has higher execution efficiency when the failure equipment data is large and contains much more noise.