【作者】 樊绍胜；

【导师】 王耀南；

【作者基本信息】 湖南大学 ， 控制理论与控制工程， 2006， 博士

【摘要】 本文着重研究冷凝器污垢清洗的智能测量与控制方法，合理、高效地实现冷凝器污垢的在线清洗，改善冷凝器传热效果，提高汽轮机组的运行效率。论文的主要研究工作如下： 论文首先探讨了冷凝器污垢产生的原因以及对冷凝器传热性能的影响，说明污垢监测、污垢清除的必要性；总结并分析了各种常用污垢测量方法以及污垢清洗方法的优点与不足；在此基础上，指出将智能技术应用于污垢测量、污垢预测以及污垢清洗的优化与控制是改善冷凝器传热性能的发展方向。 针对冷凝器在线清洗的具体要求，综合常规清洗方法的优点，提出了一种冷凝器在线清洗的新方案。根据此方案，研发了清洗控制系统的软、硬件。 提出了基于特征选取的冷凝器污垢智能测量新方法。该方法采用基于BP神经网络的灵敏度计算，从受污垢影响的冷凝器各个性能指标中，提取最能反映冷凝器污垢状态的特征变量；在此基础上，针对冷凝器变工况、冷凝器空气漏入量等因素对污垢特征变量的影响，研究基于T-S模型、多RBF神经网络、对角递归神经网络的智能建模方法，有效实现冷凝器污垢与其他参数变化对特征变量影响的分离。根据此方法，进行了现场试验，试验结果表明：该方法能较准确地在线监测冷凝器污垢，并在冷凝器出现堵管或空气漏入量较大时，取得比热阻法、传热系数法更可靠的测量结果。 根据冷凝器运行时污垢积聚的特点，分析了冷凝器的周期性结垢过程，提出了一种冷凝器污垢预测的新方法。从试验结果得出如下结论：冷凝器污垢的智能预测方法能有效处理周期性结垢过程的残余污垢现象，取得比常规模型更好的预测结果。 由于两机器臂清洗喷枪是一种高度非线性和强耦合的控制对象，且在水中工作时，具有诸如摩擦、负载变化等不确定因素，使得高精度、稳定快速的轨迹跟踪控制变得较为困难。为此，论文研究了清洗喷枪的自适应模糊控制方法、滑模控制方法、神经网络滑模控制方法。清洗喷枪轨迹跟踪的仿真结果表明，上述方法具有很强的抗干扰能力和很好的动态特性，而且当系统参数改变时，仍然有较好的响应性能。 提出将清洗液浓度的在线优化与清洗液浓度的动态控制相结合的污垢清洗控制方法，即：通过基于改进GA的多目标优化算法确定最优清洗液浓度；通过更多还原

【Abstract】 For the purpose of improving the heat transfer performance of condenser and increasing the efficiency of thermal cycle of turbo-generator unit, this dissertation investigates the approaches of intelligent measurement and control for fouling cleaning in condenser. The main topics of the dissertation are described as follows.Firstly, the dissertation explains the necessity of fouling monitoring and fouling cleaning by investigating the cause of fouling accumulation and the influence on heat transfer performance of condenser, and analyzes the advantage and disadvantage of the ordinary fouling measurement and cleaning methods. Then, basic idea and main methods of intelligent technology are introduced and a conclusion is made, which demonstrates the application of intelligent technology in fouling measurement and control is an effective way for improving the heat transfer performance of condenser.A novel strategy which combines high pressure water cleaning with chemical cleaning is proposed in this dissertation. Based on the strategy, a cleaning system is developed and the detail of the control system is described.A novel approach based on feature selection and intelligent modeling for precise measurement of fouling is proposed. In the approach, sensitivity calculation based on BP neural network is introduced to determine the most suitable feature variable which reflects the fouling state. T-S model, multi-RBF Neural network and diagonal recurrent neural network are employed to separate the influences of both fouling and other factors, such as off-design condition on feature variable. Based on it, an experimental system is developed and experiment on an actual condenser is carried out. The results show the approach measures the fouling correctly, and is more effective than thermal resistance method or heat transfer coefficient method under the condition of blocked tubes or excessive amount of air in condenser.The prediction of fouling in condenser is heavily influenced by the periodic fouling processes and dynamics change of the parameters. To deal with it, a novel prediction approach based on intelligent technology is proposed. Based on it, experiments are conducted. The results show the approach achieves fouling prediction precisely and is more effective than asymptotic fouling model and improved asymptotic fouling model.更多还原