2 *. 20037 2. 50640 CUSUM EWMA PCA TS79 A DOI 0. 980 /j. issn. 0254-508X. 207. 08. 004 Incipient Fault Detection in Papermaking Wastewater Treatment Processes WANG Ling-song MA Pu-fan YE Feng-ying XIONG Zhi-xin ZHAO Xiao-yan LIU Hong-bin 2 *. Jiangsu Provincial Key Lab of Pulp and Paper Science and Technology Nanjing Forestry University Nanjing Jiangsu Province 20037 2. State Key Lab of Pulp and Paper Engineering South China University of Technology Guangzhou Guangdong Province 50640 * E-mail hbinjm@ 63. com Abstract Incipient fault detection has been an important research topic in the field of process monitoring. However traditional multivariate statistical process monitoring methods fail to detect incipient faults. In this paper two methods were developed for the incipient fault detection of a papermaking wastewater treatment process multivariate cumulative sum combined with principal component analysis PCA and multivariate exponent weighted moving average combined with PCA. The results proved the effectiveness of the proposed fault monitoring methods. Key words papermaking wastewater treatment process principal component analysis cumulative sum CUSUM exponent weighted moving average EWMA fault detection Cumulative sum CUSUM Exponent weighted moving average EWMA 6 2 CUSUM EWMA 3-5 207-04-25 2060 6305996 20530 * 20 China Pulp & Paper Vol. 36 No. 8 207
I 7 Chen Q CL 5 8 CUSUM EWMA Principal component analysis PCA Q = θ α 2 c CL 槡 2θ 2 h0 + + θ /h 0 2h 0 h 0-5 2 θ θ θ d = n j = r + λd j d = 2 3 h 0 = - CUSUM EWMA 2θ θ 3 /3θ 2 2 c α α CUSUM Multivariate CUSUM MCUSUM EWMA Multivariate EWMA MEWMA MCUSUM MEWMA. CUSUM X m n m EWMA n 0 PCA X X = t p T + t 2 p T 2 + + t r p T r + E = TP T + E MCUSUM T R m r P R n r E R m n r d PCA X m n T 2 Q t { t x x k = x k x k 2 x k n T i = i t d y t = 6 2 T 2 t x i = i = t -d+ d t m T 2 k PCA = x T k PΛ - P T x T k 2 d Λ = diag λ λ 2 λ r λ X T 2 CL F 3 CL T 2 = r m + m - m m - r F r m - r α 3 F r m - r α r m - r F. 3 MEWMA α ( ). 2 MCUSUM MCUSUM MEWMA PCA MCUSUM-PCA MEWMA-PCA d x i i m y t t d PCA PCA PCA Q EWMA 4 Q k PCA = x T k I - PP T x k 4 MEWMA PCA 207 36 8 2
8 EWMA MEWMA COD inf ph 7 COD inf 20% z k = ωx k + - ω z k - 0 < ω z 0 = 0 7 X R m n MEWMA Z R m n Z S Z ω / 2 - ω S S X 00 70 MEWMA PCA PCA 2 T 2 8 8 T 2 k MEWMA-PCA = z T k P ( ) ω Λ 2 - ω - P T z k 8 Q 9 8 Q k MEWMA-PCA = z T k I - PP T z k 9 2 2. 70 8 COD inf COD eff SS inf SS eff Q 0 4 m 3 /d ph T DO mg /L 9 0 ph 0. 05 2 MAT- LAB 70 COD inf t + 48 ph t + 0. 05 t 2 2. 3 MCUSUM MEWMA 2. 3. MCUSUM d MCUSUM d MCUSUM-PCA d 2. 2 d d d 2 22 China Pulp & Paper Vol. 36 No. 8 207
2. 3. 2 MEWMA λ MEWMA ω MEWMA- ω PCA ω ω = 0. 2 2. 4 2. 4. ω 0 MEWMA PCA MCUSUM-PCA MEWMA-PCA 3 ~ 5 3 PCA 4 MCUSUM-PCA d = 2 5 MEWMA-PCA ω = 0. 2 207 36 8 23
PCA MCUSUM-PCA MEWMA-PCA 2 MCUSUM-PCA MEWMA- PCA Q 0. 96 0. 94 MCUSUM-PCA 2 PCA MCUSUM-PCA MEWMA-PCA MEWMA-PCA 2 ~ 70 3 ~ 4 4 5 PCA MCUSUM- MEWMA- PCA MCUSUM- MEWMA- PCA PCA PCA PCA PCA 3 T 2 0 0. 84 0. 8 0. 32 0. 52 0. 42 Q 0. 6 0. 96 0. 94 0 0. 76 0. 72 6 PCA 7 MCUSUM-PCA d = 2 8 MEWMA-PCA ω = 0. 2 24 China Pulp & Paper Vol. 36 No. 8 207
MEWMA-PCA MCUSUM-PCA 4 5 PCA PCA Q 0. 6 PCA 2. 4. 2 PCA MCUSUM-PCA MEWMA-PCA 6 ~ 8 MCUSUM-PCA MEWMA-PCA 2 ~ 70 0 ~ 5 7 8 PCA 6 PCA MCUSUM-PCA MEWMA-PCA 2 MCUSUM-PCA Q 0. 76 MEWMA-PCA Q 0. 72 diagnosis J. Journal of Shanghai Jiao- based on multivariates statistical analysis J tong University 205 49 6 842. J. 205 49 6 842. fault diagnosis methods J 29 2 57. PCA Q 0 PCA 202 29 2 57. Q T 2 J PCA MCUSUM-PCA MEWMA-PCA T 2 Q 3. Industrial & Engineering Chemistry Re- MCUSUM process monitoring J search 200 40 6 56. MEWMA PAC MCUSUM-PCA MEWMA-PCA PCA China Pulp & Paper 206 35 0 3. MCUSUM-PCA MEWMA-PCA PCA J MCUSUM-PCA MEWMA-PCA PCA MCUSUM-PCA MEWMA-PCA 2 ~ 3 0 ~ 5 HUANG Dao-ping QIU Yu LIU Yi-qi et al. Review of data-driven fault diagnosis and prognosis for wastewater treatment J. Journal of South China University of Technology Natural Science Edition 205 43 3.. J. 205 43 3. 2 ZHOU Dong-hua HU Yan-yan. Fault diagnosis techniques for dynamic systems J. Acta Automatica Sinica 2009 35 6 748.. J. 2009 35 6 748. 3 Liu Tian-long Shen Wen-hao. A review of applications of fault diagnostic expertsystem in wastewater treatment J. Paper Science & Technology 20 30 2 75.. J. 20 30 2 75. 4 QIN S. Joe. Survey on data-driven industrial process monitoring and. Annual Reviews in Control 202 36 2 220. 5 JI Hong-quan HE Xiao ZHOU Dong-hua. Fault detection techniques. 6 LI Juan ZHOU Dong-hua SI Xiao-sheng et al. Review of incipient. Control Theory & Applications 202. J. 7 GE Zhi-qiang YANG Chun-jie SONG Zhi-huan. Research and application of small shifts detection method based on MEWMA-PCA. Information and Control 2007 36 5 650.. MEWMA-PCA J. 2007 36 5 650. 8 CHEN Junghui LIAO Chien-Mao LIN Franz Ren Jen et al. Principle component analysis based control charts with memory effect for 9 YANG Hao MO Wei-lin XIONG Zhi-xin et al. Soft Sensor Modeling of Papermaking Effluent Treatment Processes Using RPLS J.. RPLS. 206 35 0 3. 0 LIU Hongbin KIM MinJeong Kang OnYu et al. Sensor validation for monitoring indoor air quality in a subway station J. Indoor and Built Environment 202 2 205. CPP 207 36 8 25