A general framework for disturbance compensation in airborne and seaborne gravimetry supported by machine learning
High-accuracy measurement systems operating in complex operational environments, such as airborne and seaborne gravimetry, are severely affected by interacting external influences (disturbances) including temperature variations, inertial accelerations, and platform rotations. This work aims to develop and prove a general method for compensating such disturbances beyond the limits of conventional approaches. We present a general framework based on three pillars: a multi-sensor system, supervised machine learning, and a dedicated laboratory -training platform-. This measurement technique was investigated through two experimental case studies relevant to airborne and seaborne gravimetry: (i) temperature and thermal-gradient rejection in a high-sensitivity tri-axial accelerometer, and (ii) pitch and roll rejection for vertical acceleration estimation. The experiments highlighted that, although the general framework provides a useful guideline, its application to airborne and seaborne gravimetry is challenging and not straightforward, requiring the development of dedicated and innovative experimental solutions. This difficulty arises from the specific nature of the measurand-gravity-which cannot be easily varied under controlled conditions. In this work, we present a dedicated approach to addressing these challenges.