Thrust allocation is of great importance for the application of Dynamic Positioning System (DPS). For dynamically positioned vessels, the thrust allocation is formulated as a nonlinear optimization problem, where the demanded forces and moments are distributed among the available thrusters. Both hydrodynamic interaction effects and physical limitations of thrusters affect the thrust generation. Therefore, the thrust allocation algorithm can be improved if these effects are considered. We propose a bivariate thrust efficiency function, dealing with both the forward thruster angle and the rear thruster angle, to describe the thrust loss. The thrust efficiency function is obtained from the model tests and approximated by the Radial Basis Function (RBF) neural network. The consequent thrust allocation problem is solved by the Sequential Quadratic Programming (SQP) algorithm with slack variables. The numerical simulations demonstrate a maximum power reduction of 16.03% compared with the forbidden zone algorithm. The proposed algorithm can also enhance system stability, highlighting the advantages of taking bivariate thrust efficiency function into account.

推力分配对于动态定位系统（DPS）的应用非常重要。对于动态定位的船舶，推力分配公式化为非线性优化问题，其中所需的力和力矩在可用推力器之间分配。流体动力相互作用效应和推进器的物理局限性都会影响推力的产生。因此，如果考虑到这些影响，则可以改进推力分配算法。我们提出了一个双变量推力效率函数，该函数同时处理前推力器角度和后推力器角度，以描述推力损失。推力效率函数从模型测试中获得，并通过径向基函数（RBF）神经网络进行近似。因此，推力分配问题通过具有松弛变量的顺序二次规划（SQP）算法解决。数值模拟表明，与禁区算法相比，最大功耗降低了16.03％。所提出的算法还可以增强系统稳定性，突出了考虑双变量推力效率函数的优势。