Development of Active Heave Compensator

升沉补偿器是用来隔离悬浮在浮体上的仪器和结构物的升沉运动。在深海中，垂荡运动的检测是非常困难的，因为不可能用超声波等直接、实时地测量。本研究开发了一种由加速度计和执行器组成的主动式运动补偿系统。该系统利用浮体平均位置保持平均水位的知识去除漂移，首先设计FIR数字滤波器对加速度进行双积分，去除漂移、高频和低频噪声。计算是在微处理器中以数字方式进行的。在测试过程中发现，浮体位置的计算对信号的任何虚假波动都是非常敏感的，例如与真实信号的低幅度稳态和很低的频率偏移。通过旁路低通滤波，检测出噪声的低频分量，并成功地从信号中去除。该滤波器可以检测带宽相对较窄的垂荡运动。其次，针对滤波精度更高的问题，提出了一种基于序贯最小二乘算法的新型滤波算法。该算法能够在不损失实时性的情况下对噪声分量进行识别。为了保证算法的永久性工作，防止参数发散，还采用了周期性重置的方法。该算法在没有漂移和相位延迟的情况下，准确地检测出了垂荡运动。该算法在起伏过程中启动时也是稳定的。在伺服控制的摆动工作台上进行了升沉补偿器的实验，验证了该算法的有效性。

Heave compensator is used to isolate instruments and structures, which are suspended from the floating body, from heave motion. In deepsea, heave motion detection is very difficult because the direct and realtime measurement by ultra sound and so on are impossible. In this research an active motion compensator system, which consists of accelerometer and actuator, has been developed. In this system, drift is removed by using the knowledge that average pisition of floating body stays mean water level.Firstly FIR digital filter was designed to double integrate the acceleration and remove the drift, high frequency and low frequency noise. The calculation is done numerically in a microprocessor. It became obvious during testing that the calculation of floating body position is extremely sesitive to any false fluctuation of signal, such as low-amplitude steady-state and very low frequency offsets from the true signal. With bypassed Low Pass Filter, low frequency component of noise is detected and successfully removed from the signal. This filter can detect heave motion of relatively narrow bandwidth. Simulation and experiment were performed and the function of the filter was confirmed.Secondly aiming more precise filter, another type of filter with sequential least square algorithm was developed. This algorithm enable to identify noise component without loss of realtimeness. Periodical reset is also employed to ensure the algorithm works permanently and to prevent divergence of parameters. With this algorithm heave motion is accurately detected without drift and phase delay. This algorithm also stable when initiated during the heaving. This algorithm was successfully tested by experiment with heave compensator on the servo-controlled oscillating stage.