基于Change-Point的风电数据挖掘算法研究

WIND TURBINE DATA MINING ALOGORITHM BASED ON CHANGE-POINT RESEARCH

风电机组运行过程中,一些故障导致设备状态发生改变,状态的改变发生在一个持续的时间序列中,找到变化点的时间对于故障回溯及根本原因分析具有重要价值。该文研究风电信号及状态时序变化的特点,引入统计学中的Change-Point算法,通过划分不同置信区间求取置信度方法解决奇异变点的不确定度问题。通过实验对算法进行验证,得出以下结论:Change-Point算法能够有效挖掘到历史数据中的一维及二维模型数据的变化,并给出变点;Change-Point算法思想是挖掘出数据本身的规律性,不受其他条件限制,因此可广泛应用于风电机组数据采集与监视控制(supervisory control and data acquisition,SCADA)系统变量数据挖掘中的问题回溯,快速定位SCADA数据状态变化点。

In the wind turbine operation,faults lead to turbine state changes,which occur in a continuous time series. To find out the change timing is of great value to fault backtracking and root cause analysis. We studied the characteristics of time series changes of wind power signals and states,and introduced a statistics algorithm Change-Point,which solved the uncertainty of singularities by calculating the confidence levels in different divided confidence intervals. We carried out experiments to verify the algorithm and concluded that the Change-Point algorithm can effectively mine the changes of the one-dimensional and two-dimensional model data in the historical data and illustrate the change points. The idea of ChangePoint algorithm is to mine the regularity from the data,without limitation from other conditions. Therefore the algorithm can be widely applied to the fault backtracking in data mining of the system variables from wind turbines supervisory control and data acquisition(SCADA)system,and applied to rapidly positioning the SCADA data state change points.