摘要
对多要素不确定性泵站故障进行有效检测,引入深度学习理论知识,提出了一种基于自稀疏编码的支持向量机泵站故障检测方法。结果表明:对于有限样本的情况下稀疏自编码的SVM方法泵站故障检测可获得较高的精度,并验证了方法的可行性、正确性及有效性;通过对比SVM、稀疏自编码的SVM、BP神经网络3种泵站故障检测方法,得出稀疏自编码的SVM在泵站故障检测中的优越性。为泵站故障检测提供了一种新的解决方法。
In order to effectively detect the failure of multi-factor uncertainty pumping station,the knowledge of deep learning theory is introduced,and a fault detection method based on self-sparse coding for support vector machine pumping station is proposed.The results show that:For the case of finite samples,the SVM method for pumping station fault detection can obtain higher accuracy,and the feasibility,correctness and effectiveness of the method are verified.By comparing the three pump station fault detection methods of SVM,sparse self-encoding SVM and BP neural network,the superiority of sparse self-encoding SVM in pump station fault detection is obtained.It provides a new solution for pump station fault detection.
作者
殷振兴
YIN Zhen-xing(New Huaishu River Management Division of Jiangsu Province,Huai’an 223005,China)
出处
《水科学与工程技术》
2019年第6期52-55,共4页
Water Sciences and Engineering Technology
关键词
泵站
故障检测
自编码
支持向量机
pump station
fault detection
self-encoding
support vector machine
