论文部分内容阅读
船舶蒸汽动力装置中除氧器压力和除氧器水位互相关联,具有很强的耦合性,传统的PID控制很难获得令人满意的控制效果,因此必须采取相应的解耦措施。PID型神经网络不仅具有传统PID的优点,还具有神经网络的自学习和逼近任意函数的能力。本研究建立了除氧器压力和水位的模型,并通过建立与比例、积分和微分相对应的神经元,将PID和神经网络整合在一起,提出一种PID型神经网络解耦控制方法。在所建立的除氧器压力和水位模型上对PID型神经网络解耦控制方法进行仿真。仿真结果表明,相对于单回路PID控制方法,该方法具有比单回路PID控制更好的解耦效果,可以将除氧器压力和水位的稳定时间分别缩短100s和60 s,并将二者的超调量分别减少0.6 KPa和0.005 m。
The deaerator pressure and deaerator water level in marine steam power plant are highly correlated. The traditional PID control is difficult to obtain satisfactory control effect, therefore, corresponding decoupling measures must be taken. PID neural network not only has the advantages of the traditional PID, but also has the ability of neural network self-learning and approximation of any function. In this study, a model of deaerator pressure and water level was established, and a PID neural network decoupling control method was proposed by integrating PID and neural network by establishing neurons corresponding to proportion, integral and differential. The PID neural network decoupling control method was simulated on the established model of deaerator pressure and water level. The simulation results show that this method has a better decoupling effect than single-loop PID control, and can shorten the settling time of deaerator pressure and water level by 100 s and 60 s, respectively, compared with single-loop PID control method. Overshoot decreased 0.6 KPa and 0.005 m respectively.