An efficient surrogate model-based method for deep-towed seismic system optimization

Abstract

The towing speed, cable length, towing depth, and hitch position should be optimized to improve the operational safety and efficiency of a deep-towed seismic system. However, the optimization is ineffective due to system nonlinearity and large-scale calculations. Therefore, a surrogate model-based method is proposed to simplify a deep-towed system and promote system optimization. The relationships among the towing speed, viscous drag and pitch angle of the towed vehicle are expressed using a surrogate model generated by using the artificial neural network method. Input variables of the surrogate model are related to the optimization variables through proper design variable selection, which generates an efficient explicit formulation. Additionally, a quasi-static iteration is established to speed up calculating the cable shape of the towing cable, and analytical models simplify the calculations of drag forces acting on the drogue and seismic array. Finally, the best design, such as the towing speed, cable length and hitch position, are obtained with limited cable tension and towing depth in a case study. This research provides an effective method for deep-towed seismic system optimization.