Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SEWAGE TREATMENT PROCESS FAULT MONITORING METHOD BASED ON FUZZY WIDTH ADAPTIVE LEARNING MODEL
Document Type and Number:
WIPO Patent Application WO/2021/109180
Kind Code:
A1
Abstract:
A sewage treatment process fault monitoring method based on a fuzzy width adaptive learning model. The method comprises the following two stages: "offline network training" and "online monitoring". The "offline modeling" comprises: firstly using a batch of normal data and four batches of fault data as training samples, and performing offline training on a network; labeling the data, that is, a normal moment is represented by zero and a fault moment is represented by one; and after network training is completed, obtaining weight parameters in order to perform "online monitoring". The "online monitoring" comprises: using newly collected data as test data, and performing online monitoring by following the same steps as that of the offline network training. With regard to output results of online monitoring, a one-hot encoding mode is used to realize zero-one discrimination of the output results of the online monitoring, wherein if the result is zero, this indicates that the running state is normal, and if the result is one, this indicates that there is a fault occurring during the running process, such that fault monitoring is realized. When comparing the present method with a method using a deep network, only the number of enhancement nodes needs to be increased, reconstruction can be performed in an incremental manner, and there is no need to retrain the whole network from the beginning. Accordingly, network training can be completed in a short time, rapid fault monitoring can be realized in a timely manner, and the method has a higher practical application value.

Inventors:
CHANG PENG (CN)
DING CHUNHAO (CN)
LU RUIWEI (CN)
LI ZEYU (CN)
WANG KAI (CN)
Application Number:
PCT/CN2019/124479
Publication Date:
June 10, 2021
Filing Date:
December 11, 2019
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
UNIV BEIJING TECHNOLOGY (CN)
International Classes:
G05B13/04
Foreign References:
CN110232256A2019-09-13
CN108628291A2018-10-09
CN106802563A2017-06-06
CN105740619A2016-07-06
JP3301428B22002-07-15
Attorney, Agent or Firm:
BEIJING SIHAI TIANDA INTELLECTUAL PROPERTY AGENCY LTD. (CN)
Download PDF: