Improving Wind Estimation Accuracy Using Model Based Techniques

摘要

Obtaining an accurate measurement of the current wind conditions is a critical part of conducting parachute research. For large-scale gliding systems such as parafoils, the wind state is observable by comparing the output behavior from well-understood scripted maneuvers with the measured behavior. For circular, cruciform and other simple canopy shapes, the glide capabilities may be so minimal that the wind state is lost within the noise in the measured canopy behavior. However, recent research into low-cost precision delivery solutions requires some way of measuring and estimating the glide performance of traditionally unguided or non-gliding canopies. Sensor packages which incorporate micro-electro-mechanical systems (MEMS) rate gyroscopes and accelerometers, barometric altimeters and GPS are readily available which cost about the same as commercial-off-the-shelf GPS loggers. This work compares commercial-off-the-shelf GPS loggers against a Federated Kalman Filter fusing a barometer and an accelerometer, with the goal of improving the fidelity of the measured descent. Experimental testing was conducted using a truncated cone decelerator. Results are presented which show a Federated Kalman Filter can fuse various sensors to achieve a great improvement in fidelity of the measured descent when compared to commercial-off-the-shelf GPS loggers. The chosen sensor hardware, parachute design and sensor fusion algorithm are explained in detail.