A parallel optimization algorithm for predictive control of marine vessel

Abstract

Abstract Control of marine moving objects is a complex task, which requires considering the nonlinearity of the control object, as well as random wind-wave disturbances. One of the modern strategies is the model predictive control. The article investigates the possibilities of parallel computing technologies to solve the problem of finding the optimal sequence of control vectors in predictive control of a marine vessel. The basic concepts and mathematical model of control systems with prediction are given. Then the optimization problem is formulated. The use of SMP-systems and parallel algorithms is substantiated. Then we describe one of the most common parallel programming technologies - OpenMp. A parallel version of the conditional optimization algorithm is presented, as well as the quality functional used in the control. The results of modeling in a multi-core system with shared memory are presented. In conclusion, further directions of research, as well as possible ways to improve performance are proposed. All presented models are implemented in C++.