This paper focuses on the issue of outlier detection for time series in the process industry. Considering the characteristics of time series in process control systems, such as high non-linearity, strong noise and the special relationship between the input and output of the controlled object, a new outlier detection algorithm is proposed. The algorithm adopts an improved Radial Basis Function Network to construct the model of the controlled object and an Auto-Regression Hidden Markov Model to detect outliers. Unlike many conventional outlier detection methods, this algorithm does not need any prior data and can detect outliers accurately without preselecting the threshold. The proposed detection algorithm is validated by the application to the electrode regulator system of an arc furnace and comparison with Takeuchi’s auto-regressive model detection approach.
AlexandridisASarimveisHBafasG (2003) A new algorithm for online structure and parameter adaptation of RBF networks. Neural Networks16: 1003–1017.
2.
AllaharKNBattocchiDOrazemME (2008) Modeling of electrochemical impedance data of a magnesium-rich primer. Journal of the Electrochemical Society155(10): E143–E149.
3.
BakerLDHofmannTMcCallumAKYangY (1999) A hierarchical probabilistic model for novelty detection in text. Unpublished manuscript.
4.
BilmesJA (2006) What HMM can do. IEICE Transactions on Information and SystemsE89-D(3): 869–891.
5.
BrockwellPJDahlhausRTrindadeAA (2002) Modified Burg algorithms for multivariate subset autoregression. Technical Report 2002-015, Department of Statistics, University of Florida.
6.
BrothertonTJohnsonTChadderdonG (1998) Classification and novelty detection using linear models and a class dependent-elliptical Bassi function neural network. In: Proceedings of the International Conference on Neural Networks, Anchorage, AK.
7.
CastetJFSalehJH (2009) Satellite and satellite subsystems reliability: statistical data analysis and modeling. Reliability Engineering & System Safety94(11): 1718–1728.
8.
DasguptaDForrestS (1996) Novelty detection in time series data using ideas from immunology. In: Proceedings of the Fifth International Conference on Intelligent Systems, Reno, NV, June 1995.
9.
DeCosteDLevineMB (2000) Automated event detection in space instruments: a case study using IPEX-2 data and support vector machines. In: Proceedings of the SPIE Conference on Astronomical Telescopes and Space Instrumentation, Munich, Germany.
10.
FawcettTProvostFJ (1999) Activity monitoring: noticing interesting changes in behavior. In: Proceedings of the 5th ACM SIGDD International Conference on Knowledge Discovery and Data Mining, San Diego, CA, USA, ACM, 53–62.
HodgeVJAustinJ (2004) A survey of outlier detection methodologies. Artificial Intelligence Review22(2): 85–126.
13.
LeeH-JChoS (2006) Application of LVQ to novelty detection using outlier training data. Pattern Recognition Letters27: 1572–1579.
14.
LiJShiJJSatzD (2008) Modeling and analysis of disease and risk factors through learning Bayesian networks from observational data. Quality and Reliability Engineering International24(3): 291–302.
15.
LiuGPKadirkamanathanVBillingsSA (1998) Online identification of nonlinear systems using Volterra polynomial basis function neural networks. Neural Networks11: 1945–1967.
16.
LouH-L (1995) Implementing the Viterbi Algorithm. IEEE Signal Processing Magazine.
17.
NarendraKSParthasarathyK (1990) Identification and control of dynamical systems using neural networks. IEEE Transactions on Neural NetworksI(1): 4–27.
RabinerLR (1989) A tutorial on Hidden Markov Model and selected applications in speech recognition. Proceeding of the IEEE77(2): 257–286.
20.
RankE (2003) Application of Bayesian trained RBF networks to nonlinear time-series modeling. Signal Processing83: 1393–1410.
21.
StoicaPSelenY (2004) Model-order selection: a review of information criterion rules. IEEE Signal Processing Magazine, 36–47.
22.
TakeuchiJYamanishiK (2006) A unifying framework for detection outliers and change point from time series. IEEE Transactions on Knowledge and Data Engineering18(4): 482–492.
23.
TheofilouDSteuberVSchutterED (2003) Novelty detection in Kohonen-like network with a long-term depression learning rule. Neurocomputing52: 411–417.
24.
TrindadeAA (2003) Implementing modified Burg algorithms in multivariate subset autoregressive modeling. Department of Statistics, University of Florida.
25.
YanWWShaoHHWangXF (2004) Soft sensing modeling based on support vector machine and Bayesian model selection. Computers & Chemical Engineering28(8): 1489–1494.
