摘要
提出了一种在线测量冷凝器污脏程度的新方法。该方法选取传热端差作为研究对象 ,以神经网络建模技术为基础成功实现冷凝器污脏、工况参数变化对传热端差影响的分离 ,可较准确地实现冷凝器污脏的在线监测。在神经网络建模中 ,采用RBF神经网络描述变工况传热端差变化的非线性过程 ,研究了一种自适应训练算法动态调整网络结构与参数 ,从而获得了结构紧凑、精度较高的测量模型 ,便于实时应用。根据此方法 ,研制了以DSP为核心的测量仪 ,并在不同工况和堵管情况下进行了现场试验 ,试验结果证明了该方法的有效性。
A novel approach for on-line measuring fouling in condenser is proposed. In the approach, terminal temperature difference is chosen to reflect the fouling state, Neural network modeling is applied to separate the influences of both the fouling and off-design condition on terminal temperature difference. During modeling, RBF neural network is employed to depict off-design condition terminal temperature difference, an adaptive learning algorithm is proposed to determine the optimum structure and parameters of neural network, this makes the measurement model compact and accurate. Based on it, an instrument is developed and test on an actual condenser under various condition and blocking tubes is conducted. The results prove the approach to be effective.
出处
《传感技术学报》
CAS
CSCD
2004年第3期431-435,共5页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金项目资助 (6 0 0 75 0 0 8)
关键词
冷凝器污脏
在线测量
传热端差
神经网络建模
fouling in condenser
on-line measurement
terminal temperature difference
neural network modeling
