SBIR Phase I: Nonlinear Modeling in the Presence of Disturbances Using Support Vector Machines

This Small Business Innovation Research Phase I project will investigate novel algorithms for developing nonlinear models based upon time series data that is affected by disturbances. The Box-Jenkins algorithm has been the standard approach over the past few decades for developing models for time-series systems that are affected by disturbances. In recent years, Support Vector Machines (SVMs) have been used to create accurate nonlinear models based upon empirical data. The proposed SBIR research will investigate combining SVM modeling approaches with Box-Jenkins type disturbance rejection techniques. Such an approach would be significantly more computationally efficient, thus, allowing commercialization of the algorithms. Because modeling of nonlinear time-series based systems that are affected by disturbances is commonly encountered across a wide variety of fields including economics, the process industries, engineering, psychology and defense, the proposed research has wide applicability.