摘要
为了解决电站燃气轮机燃烧室故障预警问题,采用一种基于多元状态估计技术(MSET)的方法。在提取故障特征量的基础上,首先利用MSET建立燃机燃烧室正常工作状态下的模型,对待观测值进行最优估计得到估计值,估计值和观测值之间的偏差可以反映燃烧室内部工作是否异常,引入偏离度定量衡量观测值与估计值之间的偏离程度,有利于捕获故障细微变化过程,然后采用滑动窗口法确定故障预警阈值,当偏离度超过故障预警阈值时,发出故障预警,提醒运行人员及时处理。用某燃气-蒸汽联合循环发电机组仿真平台的燃烧故障信息将上述算法进行了验证。结果表明:该方法可以及时发现燃烧室燃烧异常,实现了燃烧室实时故障预警,在提高机组可靠性和发电小时数的同时,缩短检修时间、降低检修成本。
In order to solve the problem of early warning of gas turbine combustion chamber failure,a multi-state estimation technique(MSET)method is adopted.On the basis of extracting the fault feature quantity,firstly,the non-parametric model of the gas turbine combustion chamber under normal working condition is established by MSET,and the estimated value of the observed value is estimated to obtain the estimated value.The deviation between the estimated value and the observed value can reflect the combustion chamber.Whether the internal work is abnormal or not,the deviation degree is quantitatively measured to determine the degree of deviation between the observed value and the estimated value,which is beneficial to capture the process of subtle change of the fault,and then uses the sliding window method to determine the fault warning threshold.When the deviation exceeds the fault warning threshold,the fault is issued.The warning reminds the operating personnel to deal with it in time.The above algorithm was verified by the combustion fault information of a gas-steam combined cycle generator set simulation platform.The results show that the method can detect the combustion abnormality of the combustion chamber in time,realize the real-time fault warning of the combustion chamber,and improve the reliability and power generation hours of the unit,shorten the inspection time and reduce the maintenance cost.
作者
黄伟
张泽发
HUANG Wei;ZHANG Ze-fa(School of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China)
出处
《汽轮机技术》
北大核心
2020年第1期38-42,共5页
Turbine Technology
基金
上海市“科技创新行动计划”地方院校能力建设专项项目(编号:19020500700)。
关键词
燃烧室
滑动窗口法
多元状态估计
故障预警
预警阈值
combustion chamber
sliding window method
multivariate state estimation
fault warning
early warning threshold
