多通道窄带Fx-Newton时频算法及动力机械有源隔振实验

提出了一种多通道窄带Fx—Newton算法,用多个控制器并联控制各窄带带通滤波器提取的多根线谱。离线辨识次级通道频响矩阵并求逆后,只需L×L个(通道数为L)2阶滤波器即可实现次级通道逆模型滤波,从而使控制器系数向最优值迅速收敛。还提出了窄带滤波相位差的自适应补偿算法,可提高线谱频率波动时的控制鲁棒性。在200kw实船柴油发电机有源无源隔振装置上进行了实验,结果表明该算法可同时控制10~11根线谱,使每根线谱衰减10~35dB。该算法运算量小,在保证收敛稳定性的同时、收敛速度大于FxLMS算法(尤其在次级通道矩阵特征值分散度大的情况)。 

MIMO Fx-Newton narrowband algorithm and experiment of active vibration isolation on diesel generator

A MIMO narrowband Fx-Newton algorithm is proposed, which can extract line-spectrums by narrowband- pass filters and control them independently by multi controllers. Frequency response matrix of secondary path is identified at the center frequency of each band and inverted oittine, and then the secondary path inverse filtering can be achieved by using L×L 2-order filters and lead controllers iterate to the optimal value rapidly and steadily. In addition, the phase shift caused by narrowband filter is compensated to improve the robustness of control system in case of frequency fluctuation. Experiments on the active-passive isolation mounting of 200 kW diesel generator show that, line-spectrums of 10-11 different frequencies can be controlled simultaneously, and the effect on each line-spectrum is 10-35 dB. The algorithm features advantages of little calculation, good stability and higher convergence rate than FxLMS algorithm especially for large eigenvalue spread of secondary path matrix.