针对某船用蒸汽动力装置主机大负荷变化时，因冷凝器水位和除氧器水位调节回路被控参数动态偏差过大影响动力装置的安全稳定运行的问题，基于传统PID控制器控制冷凝器、除氧器水位，引入单神经元控制算法，通过改进将单神经元学习所需要的状态量全部换成差值形式，在MATLAB平台上完成仿真实验。仿真实验结果表明：改进后的单神经元控制算法抗干扰能力强，响应速度快，水位被控参数超调量相比传统PID控制减小约0.1 m，调节时间缩短约200 s。

When the large load of the main engine of a certain marine steam power unit changes，the dynamic deviation of the controlled parameters of the condenser water level and the deaerator water level adjustment circuits is too large，affecting the safe and stable operation of the power device.This paper discusses the introduction of a single neuron control algorithm on the basis of the traditional PID controller to control the condenser and deaerator water level，and improves it and replaces the state parameters required by single neuron learning with the deviation form totally. The simulation experiment is completed on the MATLAB platform.Simulation experiment results show that the improved single neuron control algorithm has strong antiinterference ability，fast response speed，and the controlled parameter overshoot of the water level is reduced by about 0.1 m and the adjustment time is reduced by about 200 s compared with the traditional PID control.

TK221

中船集团自立科技项目（202109Z）；黑龙江省自然科学基金（TD2021E001）