Hybrid Control of Marine Vessels: Dynamic Positioning in Varying Conditions

Abstract

The next generation marine control systems will, as a step towards increased autonomy, have more automatic functionality in order to cope with a set of complex operations in unknown and varying environments, while maintaining safety and keeping costs low. In this thesis, a hybrid control concept for marine vessels is proposed, in order to increase the operational window of marine vessels in varying environmental conditions, with automatic switching of observers and controllers based on performance monitoring of the system states and signals, and characterizations of the sea state. The concept provides functional redundancy in the design methodology, giving better robustness to failures. The estimation and control algorithms presented in this work, are primarily designed for dynamic positioning (DP) control systems. A vessel in DP uses the thrusters to automatically maintain a fixed position, or move along a track at low speed. Vessels with DP capabilities are useful for marine operations in many marine applications, for instance; off-shore oil and gas, offshore wind, aquaculture, fisheries, rescue, ocean science, and tourism. As a result of more automatic functionality and system integration in marine control systems, control engineers must handle increasingly larger and more intricate systems. Typically, transit and maneuvering speed operation functionality can be merged with the DP functionality, giving one uni_ed system for all speed ranges, modes of operation and environmental conditions. This is a hybrid control system, with continuous-time vessel dynamics and discrete-time (automatic) switching between candidate algorithms. Performance monitoring functions decide which candidate algorithms to use in closed-loop control, detect faults, issue alarms to the operator, and provide decision support. By applying a hybrid mathematical framework for marine control system modeling, rigorous analysis of the system, including switching dynamics, can be done, and stability constraints and robustness properties may be found. In order for a hybrid control system to be reliable, good switching criteria that are robust to measurement noise and system errors need to be established for the vessel speed, use modes, and environmental conditions. Some highlights from the thesis include: • A residual calculation-based computationally efficient and reliable sea state estimation algorithm is proposed. The algorithm uses the heave, roll and pitch response spectra and simplified semi-analytical expressions for the motion transfer functions to compute an estimate of the wave spectrum. From this, characteristic periods, wave height, and the wave direction are derived. The algorithm may be used in performance monitoring functions of a hybrid controller. • Two ways of improving the transient response of a vessel in DP are investigated. Combining signal-based and model-based observers into one hybrid observer, and a time-varying model-based observer are both promising approaches. The strategies include a performance monitoring function that detects when a transient occurs. • A novel signal-based observer concept is proposed for position and velocity estimation in DP. Noisy measurements with di_erent, not necessarily periodic, sampling rates are combined into a hybrid system. The observer reacts fast to transients, and produces estimates of lower signal variance than the measurements. The proposed sea state estimation, observer and controller algorithms are tested through simulations and model-scale experiments in the Marine Cybernetics Laboratory (MCLab) at NTNU. Some of the algorithms are also tested in full-scale sea trials on the NTNU owned Research Vessel Gunnerus.