A Virtual Sensing Method for Tonal ANVC Systems

Virtual Sensing (VS) is considered to be an extension for feedforward Active Noise and Vibration Control (ANVC) systems when, for example, it is desired to minimize the response at one or more spatial locations where it is either physically impossible or undesirable to place an actual error sensor. In this context, VS is an estimation technique for predicting the appropriate system response using available measurements and a dynamic system model. A hybrid adaptive feedforward observer is proposed which has the ability to overcome the limitations of conventional dynamic observer designs. The hybrid observer utilizes a conventional dynamic observer augmented with an adaptive feedforward element for estimating the effect of the persistent disturbance. For simplicity, we restrict this development to a single tonal disturbance for which a coherent reference is assumed available. It will be appreciated that this technique may be extended to handle disturbances that contain multiple tones as well as broadband noise, as long as suitable reference signals are available. Numerical simulations and real-time experiments were performed on a one-dimensional acoustic duct. The results demonstrate that the hybrid adaptive feedforward observer is an effective method for predicting the virtual sensor response in an ANVC system.