基于粗糙径向基函数网络的船舶发电机励磁控制

Ship Generator Excitation Control System Based on Rough Set Integrated with Radial Basis Function Networks

粗糙集和神经网络的集成技术综合利用了粗糙集理论数据分析与决策规则自动提取的优点以及神经网络对非线性函数任意逼近的能力,为复杂非线性系统的建模辨识提供了一种新的途径。文中提出了一种基于粗糙径向基(radial basis function,RBF)网络的船舶发电机励磁神经比例–积分–微分(proportion-integral-differential,PID)自适应控制方法,通过粗糙RBF网络离线学习和在线辨识对神经PID控制器的参数进行自适应调节。仿真结果表明,该控制方法与传统PID控制相比具有超调量小、调节速度快等优点。

The integration of rough set with neural network comprehensively can utilize the merits of rough set, such as data intelligent analysis and automatic extraction of decision-making rules, and the ability of arbitrarily approximating nonlinear functions in neural network, so it offers a new approach for modeling and identification of complex nonlinear system. Based on the integrated of rough set with radial basis function （RBF） neural networks, the authors propose an adaptive neural P1D control method for excitation system of ship synchronous generator, which adaptively regulate the parameters of neural P1D controller by means of off-line study with rough-RBF network and online identification. Simulation results show that comparing with traditional PID control methods, the regulation speed of proposed control method is faster and the overshoot is small.