Model Order Reduction Technique to Aid Control System Design 2021-26-0347

Design of real time active controls for structural dynamics problems requires a very precise mathematical model, to closely determine the system dynamic behavior, under virtual simulation. The finite element models can somehow be used as a mathematical model but due to complex shape/structure of the component, the size of discrete models resulting from finite element analysis is usually very large, causing the virtual simulation to be extremely computationally intensive and time consuming, also the boundary conditions applied are not very scalable, making the system deviate from its real dynamic behavior. Thus, this paper deals with the design of a Model Order Reduction technique, using orthogonal decomposition of system matrices, which can be used for creating accurate low-order dynamic model with scalable boundary conditions.

The technique works in 3 phases namely, extraction of system matrices from software tools (such as ANSYS), Development of a second order reduced model using object-oriented programming language (such as MATLAB), and Deployment of model in form of state space matrices for model-based Design software (such as Simulink)

This paper presents two case studies, first one done for a simplistic beam structure (cantilever) where the model is reduced using the MOR technique and the results (accelerations/displacements and mode frequencies) are validated with experimental and theoretical results. Second case study is done on an actual automotive component (Handle-bar) under an actual problem statement.