When multiple sources contribute competitively to the noise level, multi-channel control architecture is needed, leading to more cost and time for control computation. We, hence, are concerned with a single-channel control method with a single-reference signal obtained from a linear
combination of the multiple source signals. First, we selected 3 source signal sensors for the reference signals and the error sensor, selected a proper actuator and designed the controllers: 3 cases of single-channel feedforward controllers with a single-reference signal respectively from
the source signals, a multi-channel feedforward controller with the reference signals from the source signals, and the proposed controller with the reference signal from weighted sum of the source signals. The weighting factors and the filter coefficients of the controller were determined
by the FxLMS algorithm. An experiment was then performed to confirm the effectiveness of the proposed method comparing the control performance with other methods for a tower air conditioner. The overall sound pressure level (SPL) detected by the error sensor is compared to evaluate their performance.
The reduction in the overall SPL was obtained by 4.74 dB, 1.96 dB and 6.62 dB, respectively, when using each of the 3 reference signals. Also, the overall SPL was reduced by 7.12 dB when using the multi-reference controller and by 7.66 dB when using the proposed controller. Conclusively, under
the multiple source contribution, a single-channel feed forward controller with the reference signal from a weighted sum of the source signals works well with lower cost than multi-channel feedforward controller.
On the expected diameter, width, and complexity of a stochastic convex hull
