摘要
针对火电机组热力参数传感器中多故障模式识别问题,同时为了克服参数随机性和离群点给ELM隐含层输出矩阵带来的病态现象,提出了一种基于拉普拉斯特征映射(LE)和极限学习机(ELM)的热力参数传感器故障诊断模型。该方法首先利用LE算法提取传感器的故障特征参量,并保持高维数据的距离相似性,减少离群点的出现,接着利用ELM作为故障模式的训练和诊断分类器。最后通过热力参数传感器的故障诊断实例验证了方法的有效性,实验结果表明,采用LE-ELM的组合进行热力参数传感器故障诊断相比其他方法具有更高的诊断准确率。
Aiming at problems of multiple fault patterns recognition for thermal parameter sensor in power generating units,also for purpose of overcoming the pathological phenomena of the hidden layer output matrix caused by parameter randomness and outliers,this paper proposed a fault diagnosis model for thermal parameter sensor based on LE algorithm and ELM. Firstly,the fault feature parameters of thermal parameter sensor were extracted by LE algorithm to keep the distance similarity of the high-dimensional data and reduce the occurrence of outliers. Then ELM was used to train and diagnose the different fault modes of thermal parameter sensor. Finally,a simulated fault diagnosis example of thermal parameter sensor was applied to demonstrate the effectiveness of our proposed method. The experimental results show that the combination of LE-ELM can achieve a higher diagnosis accuracy compared with other methods referred in this paper.
出处
《仪表技术与传感器》
CSCD
北大核心
2017年第10期93-97,共5页
Instrument Technique and Sensor
基金
国家自然科学基金项目(61663013
10461026)
江西省自然科学基金项目(20161BAB212051)
江西省重点研发计划项目(20161BBE50076)
江西省教育厅科学技术项目(GJJ160491)
