Improved EMD-based Oscillation Detection for Mechatronic Closed-Loop Systems

摘要

A key aspect of control development for mechatronic systems is the simulative verification. Extensive simulation studies allow early-stage adaptations and quality and safety assessments prior to validation on the real system. The verification of modern systems generates a huge amount of data that requires automated verification methods. A crucial part of the verification is the detection of oscillations, which can deteriorate the closed-loop performance and in the worst case even lead to destabilisation. In this paper, we present an oscillation detection method that derives from the continuous process plant monitoring and is suitable for many control problems in the domain of mechatronic systems. A core element is the Empirical Mode Decomposition (EMD) combined with the assessment of the decomposed modes. For high robustness and computational efficiency, the method is extended by a tailored signal pre-processing. A normalised oscillation index provides a measure for the oscillation strength and enables the systematic analysis of crucial control signals to reveal possible optimisation potentials. A multi-component control loop for a complex nonlinear mechatronic system in the development stage is examined to demonstrate the applicability of the proposed method. The results show the reliable detection and analysis of crucial oscillations. The method is used for quality improvement and assurance already in the development phase of a product or system.