一种基于海量数据挖掘的设备状态预测算法

提出了一种基于海量数据挖掘的设备状态预警算法。工业设备有大量的历史运行数据，并且实时采样的数据维度多，数据量大，算法首先对设备良好运行状态下的大量历史数据进行自适应聚类分析，建立设备的数学模型，并根据此类模型和设备运行的实时状态值对设备的运行状态进行预测。该算法充分考虑工业应用的实际需求，自动确定聚类的数目，解决了传统聚类算法处理海量历史数据时的开销大和效率低的问题，并且保证了回归预测过程的高效性。仿真实验表明，该算法能够有效地处理海量数据，并且能够实时得到预测值，实现对设备的实时监控预测。

Equipment Condition Monitoring Algorithm Based on Massive Data Mining

An ectuipment condition monitoring algorithm based on massive data mining was proposed. Industrial equipment has massive historical data and has a lot of multidimensional real-time running data. The proposed algorithm makes adaptive cluster analysis with massive historical healthy data to establish the mathematical models of equipment.The algorithm combines these models and real-time running data to achieve predication data. This algorithm can automatically determine the count of clusters by fully considering actual requests from industrial applications,which solves the problem that traditional clustering algorithms have much spending and low efficiency, and it also guarantees the efficiency in the procedure of regression. Simulation results show that the algorithm can effectively deal with massive data and get real-time predicted values,which realizes equipment condition monitoring.