采用核鲁棒非负矩阵分解的执行器动态过程故障检测

Actuator dynamic process fault detection using kernel robust non-negative matrix factorization

中文关键词  核鲁棒非负矩阵分解  执行器  动态调节  故障检测  SPE统计量

英文关键词  kernel robust non-negative matrix factorization  actuator  dynamic regulation: fault detection  SPE statistic

中文摘要

      作为流程工业中控制系统的重要组成部分, 执行器的正常工作对确保生产过程的安全性和可靠性具有重 要意义. 针对处于动态调节中的执行器, 提出一种基于核鲁棒非负矩阵分解(KRNMF)的故障检测方法. 首先, 利用 正常状态下的历史数据构建流量特性曲线, 获取完备的动态工况训练集. 其次, 为克服运行数据动态特征、非线性 特征, 在核非负矩阵分解的基础上引入稀疏误差矩阵, 隔离异常数据. 同时, 构造新的SPE统计量并使用核密度估计 确定其控制限. 通过DAMADICS仿真、水箱平台以及火电厂减温水调节阀实际数据的对比实验, 验证了该方法的有 效性.

英文摘要

      As an important part of the control system in the process industries, the proper functioning of the actuator is of great importance to ensure the safety and reliability of the production process. A fault detection method based on kernel robust non-negative matrix factorization (KRNMF) is proposed for actuators in dynamic regulation. The flow characteristic curves are constructed by using historical data under normal conditions to obtain a complete training set of dynamic operating conditions. To overcome the dynamic and non-linear characteristics of the operating data, a sparse error matrix is introduced on the basis of the kernel non-negative matrix factorization (KNMF) algorithm to isolate the abnormal data. At the same time, a new SPE statistic is constructed and its control limits are determined by using kernel density estimation. The effectiveness of this method is verified through comparison experiments between DAMADICS simulations, water tank platform and actual data from desuperheated water control valves in thermal power plants.