基于模式识别的小管道沸腾在线监测新方法

A new online monitoring method for boiling in small channels based on pattern recognition

为了对小管道中的沸腾进行更好地在线监测,基于模式识别理论提出一种适用于小管道沸腾的在线监测新方法。首先利用电容耦合式非接触电导测量传感器,实时获得能够反映小管道内沸腾液体等效电导信息的电压信号,其次对电压信号进行数据采集与特征提取,最后选取2种模式识别算法(感知器算法和K-means聚类算法)训练分类器,对比2种分类器对小管道沸腾过程阶段辨识效果。研究结果表明,所提出的小管道沸腾过程在线监测新方法是可行、有效的,新方法将小管道沸腾过程在线辨识分为4个阶段,辨识准确率可达90%以上。

A new online monitoring method for boiling in small channels based on pattern recognition was proposed for better online monitoring of boiling in mini-channels.Voltage signals were obtained by a capacitively coupled contactless conductivity detection sensor,which can reflect equivalent conductivity information of liquid in small channels under real time.Data acquisition and feature extraction were then carried out.Finally,two pattern recognition algorithms(perceptron algorithm and K-means clustering algorithm)were selected to train classifiers,and their identification effects on boiling in small channels were compared and analyzed.The experimental results show that the new online monitoring method is effective and feasible.It can identify boiling in small channels online as four stages,and the recognition accuracy can reach over 90%.