【作者】 姬春慧；

【导师】 赵永生；

【作者基本信息】 大连海事大学 ， 轮机工程， 2004， 硕士

【摘要】 本文以船用锅炉为被控对象建立了基于单神经元的参数自整定神经网络PID控制器和基于小脑模型与PID并行控制的控制器。论文对这两种神经网络控制器在锅炉水位和汽包压力控制回路中的应用进行了研究，对神经网络控制器的性能及稳定性进行了分析和探讨。 论文首先根据船舶锅炉的工作原理，通过机理建模的方法，得到锅炉汽包水位和汽压调节对象的模型，并通过一定的假设和简化，得到汽包水位和压力在不同扰动条件下的数学模型。 论文接着讨论了神经网络中单神经元控制算法和小脑模型(CMAC)的基本原理、基本特点及运行机理。通过仿真，以锅炉水位和汽压为对象对单神经元控制器和CMAC控制器性能进行了研究，并与常规PID进行了比较和分析。仿真结果表明，单神经元控制器和CMAC控制器能克服常规PID控制的缺点，具有较强的鲁棒性与自适应性，学习速度快，适于在线学习控制，完全适用于锅炉控制系统。本文的控制算法都取得了令人满意的控制结果。更多还原

【Abstract】 Two on-line learning control method based on the single neuron controller and cerebellar model articulation controller (CMAC) neural network are presented for the parameter’s adaptive ability of the drum water system and stream pressure controlled loop of marine boiler.The accurate models of boiler drum water system and steam pressure controlled system were established in accord with the theory of marine boiler.Then the basic principle, characteristics and their shortcomings of single neuron and CMAC neuron are illuminated. Based on the simulation, the network’ structure is illustrated and it is compared with conventional PID. The simulation results show that the CMAC controller has very good robustness and adaptive ability. More over the method can use a high learning rate, which qualifies it to be used for learning online. The simulation results demonstrate the effectiveness and feasibility of each integrated intelligent control algorithm presented in the thesis.更多还原