基于航迹误差预测模型的船舶自适应控制

Ship self-adaptive control based on predicting tracking error model

为提高船舶控制精度,根据船舶航迹、航向、航速、舵角特性和历史数据,采用卡尔曼滤波进行误差预测估计,利用反传多层感知器自适应网络建立船舶航迹误差预测模型,并采用舵角、航向、航迹三层串级回路系统结构,完成自动舵控制功能.在风浪干扰、改变船舶模型的回转性指数、追随性指数、延迟因子和积分因子情况下,该系统以较少的舵角动作迅速收敛,减小了航迹的波动幅度和次数,使船舶航迹与预定航线更加拟合.仿真结果表明,在模型失配情况下,该系统仍可保持稳定的输出和光滑的控制作用,具有较好的鲁棒稳定性和良好的动态调节品质.

To improve the controlling precision of ships, a dynamically predictive model was established based on the parameter characteristics of ship＇s tracking error, course, speed, rudder angle as well as history data for these variables. Kalman filter and a back propagation multi-layer perceptrons adaptive neural network were also adapted into the predictive model, and a new autopilot configuration was established using 3 feedback loops in a cascade arrangement which performed a relatively optimal control function. The autopilot based on prediction model can converge quickly with less rudder actions when the turning ability index, turning lag index, time delay factor, and integrating factor are changed. The method can reduce the amplitude and frequency of rudder fluctuation and make the ship＇ s track fit the scheduled course better compared with other methods. Simulation results show that a stabilized output and a smooth control can still be hold even under the model mismatches and the load disturbance , which demonstrates good robust stability and dynamic regulating performance.