PCA weight and Johnson transformation based alarm threshold optimization in chemical processes

doi:10.1016/j.cjche.2017.10.027

摘要/Abstract

摘要： To alleviate the heavy load of massive alarm on operators, alarm threshold in chemical processes was optimized with principal component analysis (PCA) weight and Johnson transformation in this paper. First, few variables that have high PCA weight factors are chosen as key variables. Given a total alarm frequency to these variables initially, the allowed alarm number for each variable is determined according to their sampling time and weight factors. Their alarm threshold and then control limit percentage are determined successively. The control limit percentage of non-key variables is determined with 3σ method alternatively. Second, raw data are transformed into normal distribution data with Johnson function for all variables before updating their alarm thresholds via inverse transformation of obtained control limit percentage. Alarm thresholds are optimized by iterating this process until the calculated alarm frequency reaches standard level (normally one alarm per minute). Finally, variables and their alarm thresholds are visualized in parallel coordinate to depict their variation trends concisely and clearly. Case studies on a simulated industrial atmospheric-vacuum crude distillation demonstrate that the proposed alarm threshold optimization strategy can effectively reduce false alarm rate in chemical processes.

关键词: Alarm threshold, Chemical process, PCA, Johnson transformation, Variable weight

Abstract: To alleviate the heavy load of massive alarm on operators, alarm threshold in chemical processes was optimized with principal component analysis (PCA) weight and Johnson transformation in this paper. First, few variables that have high PCA weight factors are chosen as key variables. Given a total alarm frequency to these variables initially, the allowed alarm number for each variable is determined according to their sampling time and weight factors. Their alarm threshold and then control limit percentage are determined successively. The control limit percentage of non-key variables is determined with 3σ method alternatively. Second, raw data are transformed into normal distribution data with Johnson function for all variables before updating their alarm thresholds via inverse transformation of obtained control limit percentage. Alarm thresholds are optimized by iterating this process until the calculated alarm frequency reaches standard level (normally one alarm per minute). Finally, variables and their alarm thresholds are visualized in parallel coordinate to depict their variation trends concisely and clearly. Case studies on a simulated industrial atmospheric-vacuum crude distillation demonstrate that the proposed alarm threshold optimization strategy can effectively reduce false alarm rate in chemical processes.

Key words: Alarm threshold, Chemical process, PCA, Johnson transformation, Variable weight

Wende Tian, Guixin Zhang, Xiang Zhang, Yuxi Dong. PCA weight and Johnson transformation based alarm threshold optimization in chemical processes[J]. Chinese Journal of Chemical Engineering, 2018, 26(8): 1653-1661.

Wende Tian, Guixin Zhang, Xiang Zhang, Yuxi Dong. PCA weight and Johnson transformation based alarm threshold optimization in chemical processes[J]. Chin.J.Chem.Eng., 2018, 26(8): 1653-1661.

参考文献

[1] Engineering equipment and materials users' association (EEMUA), Alarm System:A Guide to Design, Management and Procurement, 2nd ed.EEMUA, London, 2007.

[2] I. Izadi, S.L. Shah, D.S. Shook, T. Chen, An introduction to alarm analysis and design, IFAC Proceedings Volumes, Spain, June 30-July 32009, pp. 645-650.

[3] J. Wang, F. Yang, T. Chen, S.L. Shah, An overview of industrial alarm systems:main causes for alarm overloading, research status, and open problems, IEEE Trans. Autom. Sci. Eng. 13(2) (2016) 1045-1061.

[4] R. Jiang, Optimization of alarm threshold and sequential inspection scheme, Reliab. Eng. Syst. Saf. 95(3) (2010) 208-215.

[5] D. Peng, X.B. Gu, Y. Xu, Q. Zhu, Integrating probabilistic signed digraph and reliability analysis for alarm signal optimization in chemical plant, J. Loss Prev. Process 33(2015) 279-288.

[6] A. Jablonski, T. Barszcz, M. Bielecka, P. Breuhaus, Modeling of probability distribution functions for automatic threshold calculation in condition monitoring systems, Measurement 46(1) (2013) 727-738.

[7] N.A. Adnan, Y. Cheng, I. Izadi, T. Chen, Study of generalized delay-timers in alarm configuration, J. Process Control 23(3) (2013) 382-395.

[8] H. Gao, Y. Xu, X. Gu, X. Lin, Q. Zhu, Systematic rationalization approach for multivariate correlated alarms based on interpretive structural modeling and Likert scale, Chin. J. Chem. Eng. 23(12) (2015) 1987-1996.

[9] E. Cuevas, F. Sencion, D. Zaldivar, M. Perez-Cisneros, H. Sossa, A multi-threshold segmentation approach based on artificial bee colony optimization, Appl. Intell. 37(3) (2012) 321-336.

[10] T. Tanatorn, N. Ekawit, T. Surapa, False alarm reduction in BSN-based cardiac monitoring using signal quality and activity type information, Sensors 15(2) (2015) 3952-3974.

[11] R.D. Fricker Jr., D. Banschbach, Optimizing biosurveillance systems that use threshold-based event detection methods, Inf. Fusion 13(2) (2012) 117-128.

[12] A. Mezache, F. Soltani, A novel threshold optimization of ML-CFAR detector in Weibull clutter using fuzzy-neural networks, Signal Process. 87(9) (2007) 2100-2110.

[13] R. Brooks, R. Thorpe, J. Wilson, A new method for defining and managing process alarms and for correcting process operation when an alarm occurs, J. Hazard. Mater. 115(1-3) (2004) 169-174.

[14] G. Verdier, N. Hilgert, J.P. Vila, Adaptive threshold computation for CUSUM-type procedures in change detection and isolation problems, Comput. Stat. Data Anal. 52(9) (2008) 4161-4174.

[15] J. Wang, H.G. Li, A class of adaptive management strategies of nuisance alarms associated with industrial process parameters, CIESC J. 66(10) (2015) 4085-4091(in Chinese).

[16] Q. Li, G.D. Clifford, Signal quality and data fusion for false alarm reduction in the intensive care unit, J. Electrocardiol. 45(6) (2012) 596-603.

[17] S. Herbold, J. Grabowski, S. Waack, Calculation and optimization of thresholds for sets of software metrics, Empir. Softw. Eng. 16(2011) 812-841.

[18] T. Chen, On reducing false alarms in multivariate statistical process control, Chem. Eng. Res. Des. 88(4) (2010) 430-436.

[19] J. Zhao, J. Zhu, Data filtering based alarm processing strategy for repeating alarms, J. Tsinghua Univ. (Sci. Technol.) 52(3) (2012) 277-281(in Chinese).

[20] W.D. Tian, X.N. Shi, C.L. Wang, C.K. Li, Alarm optimization of chemical processes based on data filtering function, Xiandai Huagong Mod. Chem. Ind. 35(4) (2015) 160-163.

[21] D.H. Xiao, H.G. Li, H. Zang, A multi-variable alarm threshold optimization method for chemical processes, Contr. Eng. China 22(2) (2015) 246-251.

[22] L. Han, H. Gao, Y. Xu, Q. Zhu, Combining FAP, MAP and correlation analysis for multivariate alarm thresholds optimization in industrial process, J. Loss Prev. Process 40(10) (2016) 471-478.

[23] H. Liu, Z. Liu, H. Li, A data-driven approach to chemical process alarm threshold optimization, CIESC J. 63(9) (2012) 2733-2738(in Chinese).

[24] L. Abdou, O. Taibaoui, A. Moumen, A.T. Ahmed, Threshold optimization in distributed OS-CFAR system by using simulated annealing technique, IEEE Sousse Tunis. 28-30(April) (2015) 295-301.

[25] H. Zang, H. Li, Optimization of process alarm thresholds:a multidimensional kernel density estimation approach, Process. Saf. Prog. 33(3) (2015) 292-298.

[26] Q.Y. Zhou, P. Tian, Z.Y. Tian, Estimation of non-normal process capability index based on Johnson system of transformations, Syst. Eng. 22(5) (2004) 98-102.

[27] K. Glendenning, T. Wischgoll, J. Harris, R. Vickery, L. Blaha, Parameter space visualization for large-scale datasets using parallel coordinate plots, J. Imaging Sci. Technol. 60(1) (2016) 104061-104068.

[28] R. Dunia, T.F. Edgar, M. Nixon, Process monitoring using principal components in parallel coordinates, AICHE J. 59(2) (2013) 445-456.