Improving the tracking performance of a two-axis marine satellite tracking TV antenna via better attitude estimation and effective testing
The marine antenna is used to track a satellite in the geosynchronous orbit. The antenna uses both signal strength peaking and active stabilization in order to acquire and track a satellite. The difficult part about the tracking was that the presence of external acceleration destabilizes the previously implemented sensor fusion. As a result, the active stabilization was compromised. In order to improve the tracking performance of the antenna, it was studied in depth in terms of how to evaluate and validate the tracking algorithms. A few methods were introduced to evaluate the operations of the antenna. A constant rate platform motion was used to test and identify the limits of the signal strength peaking mechanism. The signal strength and motor positions feedbacks to some specified constant rate motion were superimposed on an actual signal strength profile in the spatial coordinates. When part of the signal strength and motor position trajectory approaches the average signal loss baseline, the limits were then identified in terms of the maximum constant rate. A sea state platform motion was used to evaluate the performance of the active stabilization mechanism and also the overall tracking with both mechanisms. To visualize and evaluate the tracking performance, the signal strength versus the resultant of the velocity of the response was plotted. Dips in the signal strength indicated the potential weak points of the active stabilization at that particular velocity. To actually test the antenna, the implementation of a test platform was also shown. The computation of the sea wave and the motion response of the vessel, and the control of the Stewart platform were all performed on a single microcontroller. A one-dimensional sinusoidal surface wave model was used to simulate the sea state condition. The approximation of the geometric solution of the vessel i.e. the four major coordinates of bow, stern, starboard, and post, was done via a first-order Newton method. Evidently, it was very efficient, for instance, it took an average of only one iteration to obtain a solution that also satisfied a very strict error tolerance criteria. A sea state transition technique, which was simply a modified cosine function, was also introduced to smoothly interpolate between different sea states. It was demonstrated to work even for a sensorless controls system. The experimental results showed the viability of the proposed techniques for implementing an average-to-low level fidelity marine vehicle simulator that can simulate the motion response of a vessel of various dimensions in different sea state conditions. To fix and improve the active stabilization, a unique decoupled sensor fusion technique was used. Essentially, the systems of equations used to track the motions of the vessel were decoupled in yaw, pitch, and roll. And to minimize the effect of external accelerations, different levels of trusts were explored on the accelerometer measurements. Evidently, the lower trust resulted in better rejection of external accelerations. The proposed sensor fusion was demonstrated to have improved the active stabilization significantly regardless of the quality of the sensor used, or the sizes of the vessels that the antenna was installed on, or in any sea state environments.