SS-DEWMA图多目标优化及在输出传感器故障检测中的应用  

Multi-objective optimization of sum of squares double exponentially weighted moving average chart and its application in output sensor fault detection

作　　者：方一鸣[1,2] 阎淑雅 李建雄 赵晓东[1] FANG Yi-ming;YAN Shu-ya;LI Jian-xiong;ZHAO Xiao-dong(Key Lab of Industrial Computer Control Engineering of Hebei Province,Yanshan University,Qinhuangdao 066004,China;Research Center of the Ministry of Education for Intelligent Control System and Intelligent Equipment,Qinhuangdao 066004,China)

机构地区：[1]燕山大学工业计算机控制工程河北省重点实验室,河北秦皇岛066004 [2]智能控制系统与智能装备教育部工程研究中心,河北秦皇岛066004

出　　处：《控制与决策》2022年第6期1656-1664,共9页Control and Decision

基　　金：国家自然科学基金项目(61873226,61803327);河北省自然科学基金项目(F2017203304,F2019203090)。

摘　　要：针对平方和双指数加权移动平均(SS-DEWMA)图难以选取合适参数同时满足数据监控的多个指标最优的问题,提出一种SS-DEWMA图的多目标优化(MO-SS-DEWMA图)数据监控方法,并将该方法用于非线性系统传感器的故障检测.首先,采用复合嵌入式均方根容积卡尔曼滤波器(CESCKF)对系统状态进行估计,并产生残差;其次,构造残差评价(数据监控)指标漏报率(MDR)和误报率(FAR)与SS-DEWMA图的两个参数的函数,并以MDR和FAR同时最小为优化目标,利用多目标粒子群优化(MO-PSO)算法对两个参数进行离线优化,将优化后的SS-DEWMA图的输出值与阈值比较,在线检测故障,其中,采用小波分析算法削弱噪声对SS-DEWMA图的影响;最后,将所提出算法用于伺服电机驱动的连铸结晶器振动系统位移传感器故障检测中,仿真和实验结果表明,该方法能有效降低故障检测的漏报率和误报率.In view of the problem that it’s difficult for the sum of squares double exponentially weighted moving average(SS-DEWMA)chart to select appropriate parameters,which satisfy multiple optimal indexes of data monitoring,a multi-objective optimization of SS-DEWMA(MO-SS-DEWMA)chart data monitoring method is proposed and applied to sensor fault detection of nonlinear systems.Firstly,a composite embedded square-root cubature Kalman filter(CESCKF)is used to estimate the states of systems and generate residuals.Then,by constructing the function between missed detection rate(MDR)and false alarm rate(FAR)of the residual assessment(data monitoring)index and parameters of the SS-DEWMA chart,and taking the MDR and FAR as the minimum cost objective functions,a multi-objective particle swarm optimization(MO-PSO)algorithm is employed to offline optimize the control width and smoothing parameter,and the output results of the optimized SS-DEWMA chart are compared with the threshold value to detect the faults online,in which,a wavelet analysis algorithm is used in order to reduce the infiuence of noise on SS-DEWMA chart.Finally,the proposed algorithm is applied to the fault detection of the displacement sensor of the continuous casting mold system driven by servo motors.Simulated and experimental results show that the proposed method can effectively reduce the MDR and FAR.

关 键 词：传感器故障检测 平方和双指数加权移动平均图 多目标优化 卡尔曼滤波器 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]