Gömülü sistemler kullanarak gerçek zamanlı sinüzoidal titreşim engelleme

Active vibration cancellation is used in various mechanical systems to prevent undesired effects of vibration such as affecting system performance, generating damage or waste energy, and causing discomfort to the user or the environment. One of these systems using active vibration cancellation is an active mount system for car engines. This system is used to cancel the vibration created by the car engine and prevent it from affecting the chassis of the car. There are many methods proposed in the literature for this particular system. They can be grouped as adaptive, classical, and robust control methods. Adaptive control methods provide better results under low uncertainties. But their performance drop under high uncertainties and they may even become unstable. On the other hand, classical control methods are more robust to uncertainties but have worse response times. Finally, robust control methods have the best stability under high uncertainties, but lose their overall performance to gain it. In this study, we propose a novel method to work with better performance under high uncertainties. The proposed method is based on Kay's sinusoidal parameter estimation algorithm, which can be used at high noise levels. However, it is improved for embedded implementation using different optimization steps. First, a crude frequency is found using a fast Fourier transform. Then, the exact frequency is estimated around this crude frequency using Kay's frequency estimator with a selected search algorithm. This step is further improved using a two-stage search approach, predefined look-up tables, and iterative calculations. Then for real-time implementation, partial input sampling, and specific real-time phase delay calculations are performed. Frequency and phase/amplitude calculation parts are also separated for increasing the phase estimation accuracy. Finally, the proposed method is implemented on an Arm Cortex-M4 microcontroller. The loopback simulations, electrical experiments, and mechanical experiments are performed and the proposed method is compared with the state of the art in literature.