基于D_S证据理论的煤粉细度融合诊断

针对电厂煤粉细度在线测量难的问题，提出利用数据融合技术识别煤粉过细、煤粉正常、煤粉过粗。在分析与煤粉细度相关的多个运行状态参量基础上，根据历史运行数据，确定煤粉细度的典型样本，然后对各状态参量应用D—S融合规则得到结果。而对于D—S证据理论应用中基本概率赋值难以确定问题，应用正态分布曲线构造相似度函数．继而得到基本概率赋值，减少了方法的主观性。根据运行数据验证，该方法能够有效诊断煤粉细度，且具有较好的鲁棒性，具有工程实用价值。

Fusion Diagnosis of Pulverized-coal Fineness Based on a D-S Evidence Theory

With regard to the difficult problem of performing an on-line measurement of pulverized coal fineness at power plants,a datum fusion technology is proposed to discriminate whether the pulverized coal is over-fine,normal and excessively coarse.On the basis of analyzing several operation-status parameters relating to pulverized coal fineness and according to historical operational data,typical samples of pulverized coal fineness have been identified.Thereafter,the results of various status parameters can be ascertained by use of D-S fusion rules.Concerning the problem that it is difficult to determine the basic probability assignment during the use of D-S evidence theory,a normal distribution curve was utilized to construct a similarity function followed by the acquisition of the basic probability assignment,thus reducing subjectivity.As verified by the operational data,the above method could effectively diagnose the fineness of pulverized coal and is characterized by relatively good robustness and practical value in engineering applications.