Number and placement optimisation of active devices in a noise active control system

Active sound control systems have been recently used, aimed at noise suppression inside turboprop aircraft as ATR42-500 and SAAB 340 or 2000. The final aim is to substitute or, more reasonably, integrate in a next future, the classical passive methods, inside a novel design philosophy able to consider the problems related to noise aspects, since the beginning: the so-called design to noise approach. Recent studies, some performed by the authors and other colleagues inside the Brite EuRam project ASANCA II, have demonstrated the non-direct correlation between the vibration and sound field. As a consequence, systems referring to sound figures and structural actuators, should be implemented, according to the well-known ASAC (acoustostructural active control) strategy. Acoustic field may be taken into account not directly, also, as suggested by CONVIS approach (control of sound by vibration suppression). Whatever is the chosen method, the performance of the active control arrangement is strictly related to the placement of the selected actuators and sensors; as it is obvious, its overall cost is instead directly connected to the number of employed channels and their combination (regarding calculation velocity. The work here presented deals with a multi-tonal attenuation problem in a typical bare propeller aircraft fuselage: three different frequencies, corresponding to the Blade Passage Frequencies (BPF), and its first two harmonics; a two-dimensional FE representation of the fuselage section, simulating the ATR42 cabin has been investigated; a standard feedforward non-adaptive algorithm is referred to; piezoceramic actuators and microphones are implemented as active devices. The best placement of both actuators and sensors is carried out by means of GENALG, a numerical code, Genetic Algorithm-based. Its peculiarities and ability in solving the optimisation problem for a very high number of discrete possible solutions, leads to the possibility of performing a parametric analysis aimed at pointing out the best number of both actuators and sensors, contemporaneously.