A QQ-Plot and its Application to Adaptive Recursive System Parameter Estimation Nasar Aldian Ambark Shashoa 1 , Sulaiman Khalifa Yakhlef 1 , Adel Saad Emhemmed 2 , Mahmoud. Elfandi 2 1 Azzaytuna University, Faculty of Engineering, Electrical and Electronics Department, Tarhuna, Libya 2 University of Tripoli, Faculty of Engineering, Electrical and Electronics Department, Tripoli, Libya Email: dr.naser.elec@gmail.com ABSTRACT Because of the presence of sporadic high-intensity measurement noise (outliers),an adaptive algorithm for the robust estimation of parameters of linear dynamic discrete-time systems is proposed in this paper. first, the sorted data versus the normal quantiles is plotted, called QQ-plot. next the ε-contaminated normal distribution of noise is adopted. Then, a data classification procedure based on the QQ-plot approachcombined with the robustified data winsorization technique, is developed, the estimation of the unknown noise statistical parameters is solved. Moreover, an iterative procedure for estimating the contamination degree  , which originated from an ML classification, is also proposed. Thus, an ε- contaminated noise distribution is estimated and, the suboptimal maximum likelihood criterion is defined, and the system-parameter estimation problem is solved robustly, using the proposed recursive robust parameter estimation scheme. KEYWORDS QQ-plot; outliers; adaptive algorithm; parameter estimation; robustness. 1 INTRODUCTION Automatic control is a vast technological area whose central aim is to develop control strategies that improve performance when they are applied to a system or a process. The results reported thus far on control design techniques are significant from both a theoretical and a practical perspective. From the theoretical perspective, these results are presented in great depth, covering a wide variety of modern control problems, such as optimal and stochastic control, adaptive and robust control, Kaman filtering, and system identification. From the practical point of view, these results have been successfully implemented in numerous practical systems and processes, for example, in controlling temperature, pressure, and fluid level [1]. Modern industry requirements are continually increasing and, apart from advanced process control techniques, system availability, reliability and safety are becoming attributes of primary importance. It is therefore there are a growing number of papers which address fault detection techniques. Model-based approaches to fault detection in dynamic systems have been received much attention over the last decades, both in research context and in the domain of application studies on real plants. The aim of model-based fault diagnosis is to generate information about faults which have occurred in target systems using actual measurements. The model-based method is referred to as an analytical redundancy, which is low-cost compared to hardware redundancy in some safety-critical applications, provided that a model can precisely simulate the behavior of a real system. Typically, the target system is considered as a continuous-variable dynamic system, which has an input U and an output Y , the detection methods generate residuals r , parameter estimates  , which are called features [2]. Consistency checking in analytical redundancy is normally achieved through a comparison between measured signals with estimated values. The estimation is generated by a mathematical model of the considered plant. The comparison is done using the residual quantities which are computed ISBN: 978-0-9891305-4-7 ©2014 SDIWC 164