一种基于改进式SSD算法的稳态检测方法

An Improved Steady-State Detection Based on the SSD Algorithm

稳态检测对热工过程中设备性能评价、系统建模及优化、故障检测及过程辨识均具有重要意义。考虑到实际工况中监测数据分布具有不确定性,本文在SSD算法的基础上,对控制限的求取做出了改进。采用基于核函数的非参数控制限算法,用核函数拟合出监控指标的概率密度函数,然后计算出该概率密度函数下满足检验水平的控制限,并结合滑动窗口法对过程数据进行稳态检测。最后将SSD稳态检测法应用到实际过程当中,以某电厂600MW机组的给水流量数据进行稳态检测,改进后的SSD算法相比较传统而言,稳态检出率显著提高。

Steady state detection is of great significance to equipment performance evaluation, system modeling and optimization, fault detection and process identification in the courses of thermal engineering.Considering the actual condition monitoring data distribution is uncertain, the paper is based on SSD algorithm, making the control limits appropriate improvements.Using the non-parametric threshold algorithm based on the kernel density estimation to fit the probability density function of monitoring indexes, and then calculates the probability density function satisfies this test level α control limits, combined with the sliding window method to process data stability detection.Finally, the application of the actual SSD steady state detection is used into the actual process, using the feed water flow data of a power plant 600 MW unit.Compared with the traditional SSD algorithm, the improved SSD enhances the steady-detection rate significantly.