Publication Type : Conference Proceedings
Publisher : IEEE
Source : 2017 IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems, SPICES 2017, 2017
Keywords : ARMA, Chemical process, Correlated data, Data validation, Error detection, Errors, Food products, gross error detection, Process information, Serial correlation, Signal processing, Variance corrections
Campus : Bengaluru
School : School of Engineering
Department : Electronics and Communication
Year : 2017
Abstract : Data validation and reconciliation (DVR) plays key role in industries because it uses process information and statistical methods to estimate correct measurements from the observed data. DR is very effective when measurement is free from gross error. Data observed from chemical processes can be serially correlated. Serial correlation in data can arise when a process, takes time to adjust, is exposed to prolonged influences, when data is manipulated or smoothened. If serial correlation is not taken care of, then gross error detection may be inaccurate. Hence dealing serial correlation is important in gross error detection (GED) techniques. Most of the techniques implemented in gross error detection require no correlation in the measurement. But experimental data may have serial correlation. In this paper, various approaches like variance correction and pre-whitening are implemented to deal serial correlation on ARMA process and measurement test (MT) is applied to detect gross error. Results portray that MT is not an efficient method for GED and amongst the above mentioned two methods, pre-whitening with low variance is better than variance correction method.
Cite this Research Publication : Hiremath, N., Naveen Kumar, S., Surya Narayanan, N.S., Jeyanthi, R., “A study of dealing serially correlated data in GED techniques”, 2017 IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems, SPICES 2017, 2017.