Robust Parameter Estimation with Orthogonal Regression Using Interval Constraint Satisfaction Technique

Parameter estimation is a first and foremost task to design a proper mathematical model. Outliers in a data can often lead to improper parameter estimation. In present work a novel technique using interval constraint satisfaction technique is suggested for getting robust parameter estimation of a system using linear orthogonal regression. In this proposed method we used M estimator with tukey’s biweight function and modified it for orthogonal regression. Tukey’s biweight is generally computed using IRLS which has computational issues of convergence and local minima. Here, benefits of interval analysis based method is utilized to develop the proposed method. Two well-known examples of datasets are used to validate the proposed method.