基于深度信念网络的船用燃气轮机自适应控制

Adaptive Control of the Marine Gas Turbine Based on Deep Belief Network

针对船用燃气轮机非线性强、惯性大和负载多变等特点,提出1种基于深度信念网络的自适应控制器。该控制器结合了深度信念网络和传统PID控制器,利用深度信念网络对燃气轮机的转速误差信息以及负载状态提取实时特征,在线调节PID参数,并将参数传递给PID控制器,由其输出控制量。通过数字仿真验证表明:该控制器满足燃气轮机转速控制的要求,并且具有良好的自适应性,在不同工况下,能够对燃气轮机转速进行准确控制,使得系统快速响应的同时无超调量。

According to the characteristics of marinegas turbine such as strong nonlinearity,large inertia and changeable load,an adaptive controller was proposed based on deep belief network.Based on the traditional PID controller,the deep belief network was used to extract the speed fault information and load state of the gas turbine real time.It adjusted the PID parameters online and passed the parameters to the PID controller which outputs the control quantity.The results of digital simulation show that the controller meets the requirement of gas turbine speed control and has good adaptability.It can accurately control the speed of gas turbine,so that the system can respond quickly without overshoot at the same time.