基于减聚类-自适应神经模糊推理的船舶航向保持控制设计

Ship Course-keeping Control Design Based on Subtractive Clustering Adaptive Neural-Fuzzy Inference

为解决船舶在非线性和不确定性条件下的常规航向保持控制参数难以确定和性能较差的问题,提出一种基于减法聚类和神经模糊推理系统(SC-ANFIS)的船舶航向保持控制设计。基于鲁棒PID控制,借助减法聚类算法的学习能力对输入样本进行聚类分析,优化模糊量化和模糊规则,继而用神经-模糊推理的方法解决船舶的不确定性问题和非线性控制问题;同时,为避免维数灾难等问题发生,采用多维隶属度函数设计一种可在线自调整的基于SC-ANFIS的航向保持控制系统,并设计仿真试验进行对比分析。仿真试验结果表明,在存在模型参数摄动和干扰的情况下,基于SC-ANFIS的航向保持控制系统可行、有效,能取得良好的控制效果。

To improve the problem that the conventional navigation course-keeping control parameters are difficult to be determined and the performance is poor under nonlinear and uncertain conditions, a new method that adaptive neural-fuzzy inference system based on subtractive clustering(SC-ANFIS) is proposed. Based on the adaptive robust PID control model, the learning ability of the subtractive clustering algorithm is used to optimize the fuzzy quantization and fuzzy rules for the input samples. Then, neural-fuzzy inference is used to improve the ship control problems under uncertainties and nonlinear conditions. At the same time, in order to avoid dimensionality disasters and other issues, multi-dimensional membership functions are used to design a SC-ANFIS course-keeping system that can be self-adjusted online. At last, the simulation experiment and conduct comparative analysis are designed. The results show that the course-keeping design is feasible and effective, and have good control effect under the perturbation and disturbance of the model parameters.