基于流形正则化半监督学习的污水处理操作工况识别方法

发布时间：2018-06-12 03:04

  本文选题：污水处理 + 极限学习机　； 参考：《化工学报》2016年06期

【摘要】：污水处理过程容易受外界冲激扰动影响,引发污泥上浮、老化、中毒、膨胀等故障工况,导致出水水质质量差,能源消耗高等问题,如何快速准确识别污水操作工况故障至关重要。针对污水工况识别过程中现有监督学习方法未利用大量未标记数据蕴含的丰富操作工况信息,采用基于流形正则化极限学习机的半监督学习方法,监视生化污水处理过程操作运行工况。该方法在学习过程中,在标记和未标记数据输入空间构建图拉普拉斯算子,通过随机特征映射建立隐含层,在流形正则化框架下,求解隐含层和输出层之间的权重,保留随机神经网络的计算效率和泛化性能。仿真实验结果表明,基于半监督极限学习机的污水处理工况识别在准确率与可靠性方面相对优于基本极限学习机方法。
[Abstract]:The process of sewage treatment is easily affected by the disturbance of external impulse, causing problems such as sludge floating, aging, poisoning, swelling and other malfunction conditions, resulting in poor quality of effluent quality, high energy consumption and so on. It is very important to identify the fault of sewage operation condition quickly and accurately. Based on manifold regularization limit learning machine, a semi-supervised learning method based on manifold regularization is proposed to solve the problem that the existing supervised learning methods do not utilize the abundant operating condition information contained in a large number of unlabeled data in the process of sewage condition identification. Monitor the operation and operation of biochemical wastewater treatment process. In the process of learning, the graph Laplace operator is constructed in the labeled and unmarked data input space, the hidden layer is established by random feature mapping, and the weights between the hidden layer and the output layer are solved under the framework of manifold regularization. The computational efficiency and generalization performance of preserving stochastic neural networks. Simulation results show that the recognition of sewage treatment conditions based on semi-supervised extreme learning machine is better than that of basic extreme learning machine in terms of accuracy and reliability.
【作者单位】： 沈阳化工大学信息工程学院;
【基金】：国家自然科学基金项目(61203102,61573364) 辽宁省教育厅科学研究项目(L2013158,L2013272)~~
【分类号】：X703
，

本文编号：2008000