Reduced Order Metamodel Development Framework for NVH 2022-01-0219

During the design conception of an automobile, typically low-fidelity physics-based simulations are coupled with engineering judgement to define key architectural components and subsystems which limits the capability to identify NVH issues arising from systems interaction. This translates to non-optimal designs because of unexplored design opportunities and therefore, lost business efficiencies. The sparse design information available during the design conception phase limits the development of representative higher fidelity physics-based simulations. To address that restriction on design optimization opportunities, this paper introduces an alternate approach to develop reduced order predictive models using regression techniques by harnessing historical measurement and simulation data. The concept is illustrated using two driveline NVH phenomenon: axle whine and take-off shudder. Firstly, critical measurements that encode critical source dynamics are obtained which include a combination of single value and frequency-dependent series data. Secondly, after categorizing these measurements into input and response variables, the data is used to train regression models. Three unique regression modeling techniques - Gaussian Process, Gradient Boosting Machine and Discrete Random Forest - are presented in the context of their suitability to model NVH phenomena. Finally, the best performing models are validated for their prediction capabilities against measurements from new vehicle designs using residual metrics. Three case studies are presented to demonstrate this approach, two for powertrain subsystems design and one to model vehicle transfer functions to be used for systems level simulations during design conception and verification phases. The paper concludes with a preliminary set of guidelines to develop similar metamodels for other systems simulations for NVH.