船用柴油机齿条位置执行器与转速BP神经网络控制研究

Research on the Control of Rack Position Actuator and Marine Diesel Engine Speed Based on BP Neural Network

针对柴油机调速系统存在的非线性和时变性,以及传统PID控制效果不理想等特点,研究了船用柴油机转速控制优化问题。利用BP神经网络的自学习能力和自适应能力,对PID控制器的参数进行在线整定。基于BP-PID控制算法与MC9S12XEP100单片机设计了执行器位置控制器软硬件,结果表明:相对于传统PID,BP-PID具有良好的控制效果,抗干扰性和适应性更好,但同时存在运算量较大的问题。在此基础上将柴油机作为控制对象,利用dSPACE半实物仿真平台与MicroAutoBox进行算法验证,结果表明:该算法满足柴油机转速控制性能指标要求,负载突变时转速超调量小于5%,能有效降低负载变化对转速的影响,可以用于实际柴油机转速控制系统研究开发。

Owing to the non-linear and time-variantion characteristics, marine diesel engine speed control system with traditional PID controller is unsatisfied. A speed controller with self-learning and adap- tive ability was designed based on Back-Propagation （BP） Neural Network to realize on-line optimization of the PID controller parameters. Using PB-PID algorithm and MCgS12XEP100 MCU, the controller was experimentally evaluated in a marine diesel engine rack position actuator. Results demonstrated that the controller possesses better adaptability and anti-jamming capability compared to the tradition PID and PB- PID,but there is also the problem of bigger calculation losd. The algorithm is verified by hardware-in-loop simulation of dSPACE and MicroAutoBox to be able to control the overshoot less than 5 % while load varying suddenly,meeting the diesel engine speed control requirements. The algorithm can be used to research and development of diesel engine speed control system.