基于Fisher最优分割法的汽轮机组稳态运行数据提取方法研究及应用

Research and Application on Steam Turbine Steady-State Operation Data Extraction Method Based on the Fisher Optimal Partion Method

针对采用静态数学模型在线计算、分析汽轮机组性能时,模型输入需要选取稳态运行时间段内数据的工程实际问题,提出了一种基于Fisher最优分割法的汽轮机组稳态运行数据提取方法,首先根据误差函数E随给定的有限个分段数k变化的曲线,将原始样本划分为若干子样本。然后在每一个子样本中,分段数k由2开始逐步递增,根据稳态运行最短持续时长、参数波动范围等条件确定最优分段数,从而获取符合要求的稳态运行数据。实例验证了方法的有效性。

In view of adopting the static mathematical model for on-line computing and analyzing steam turbine performance, model input data need to select the steady state operation period data, steam turbine steady-state operation data extraction method based on the fisher optimal partion method is proposed. First of all, according to the error function E change with a given finite segment number K, the original sampie is divided into several sub sample. First of all, according to the error function E change with a given number of a finite number of piecewise k curve, the original sample is divided into several sub sample. Then, in the samples of each sub sample, the section number K gradually increases starting with 2, the optimal segmentation is determined according to the steady state run the shortest duration, variation of parameters, thus the steady state operation data for the requirements is obtained. Examples demonstrate the effectiveness of the method.