Real-Time Detection of Low-Frequency Components

This paper proposes a procedure for the real-time detection of low-frequency components within the spectrum of a signal related to a monitored quantity. The problem is common within different domains, such as medical diagnostics, mechanics, geology, and the monitoring of human activities. This procedure has also been specifically designed to allow the detection of more than one component, with a good compromise between selectivity and promptness. The performance of the procedure is evaluated in one specific case study: the control of a magneto-rheological suspension system of a motorbike, which requires the sudden detection of the resonance frequencies due to both the sprung mass and to the unsprung masses to absorb oscillations. In this paper, after the description of the method, some details are given about the implementation on a floating-point digital signal processor and on a fixed point microprocessor.