Most of the Real-world systems are not amicable to mathematical modelling approach due to their inherent nonlinear dynamics, unstructured uncertainties in addition to nonlinear interaction between their subsystems. Though some mathematical models exist for some systems, they fail after some levels of complexity.
As per present understanding, the best representative of the any system is the data that has been observed while the system is being operated. In the thesis, an innovative framework has been demonstrated the plausibility of finding the qualitative and quantitative behaviour in terms of time as well as parameter space using unified approach of Recurrent Neural Networks and Bifurcation & Chaos Theory. This is a tremendous advantage in terms of System Analysis, Control and Optimisation.
Key achievements of the doctoral research:
· Plausibility of the finding structure in time as well as parameter space.
· Qualitative Behaviour Analysis by constructing Bifurcation Diagram
· Quantitative Behaviour Analysis by estimating the Maximum Lyapunov Exponents in n-parameter space.
· A framework has been developed for Controlling based on the concept of OGY(Edward Ott, Celso Grebogi, and James A. Yorke) method using Recurrent Neural Networks and Genetic Algorithms approach.
The concept has been implemented on known chaotic system like Hénon map and Ikeda map. Further, it has been implemented on a Real-World Process by name Submerged Arc Furnace.