Imbalanced Data Classification Algorithm Based On Stacking And Its Application In Wastewater Treatment Fault Diagnosis

The wastewater treatment process is very complicated,including a series of physical actions and biochemical reactions,and it has the characteristics of large lag,nonlinearity and strong interference.Due to the many factors that affect the sewage treatment process,the actual wastewater treatment plant will inevitably have abnormal conditions when the system is running,and when the sewage treatment system is in a faulty operation state,it will cause serious issues such as effluent water below quality specification and secondary environmental pollution.Therefore,it is very necessary to use effective fault diagnosis technology to accurately monitor the system operation status in the sewage treatment process,timely diagnose the fault status and take corresponding recovery measures.This paper takes sewage treatment as the background,taking the imbalanced characteristics of sewage data into account.And a series of research work on sewage fault diagnosis technology based on artificial intelligence algorithm has been done in this paper.The specific research work is as follows:First,this article introduces the research background of the imbalanced data classification problem and expounds the research status of the sewage treatment fault diagnosis problem The important parameters of the sewage treatment process and common failure types are analyzed in later.At last,the characteristics and data of the imbalanced data set are described and the performance evaluation indicators are also introduced.Then,this paper proposes Stacking-based integrated wastewater treatment fault diagnosis technology,introduces the overall framework of Stacking integrated algorithm and the basic theory of four traditional classification algorithms,and determines the base classifier and meta-classifier in Stacking integrated framework through experiments;then this study selected 17 sets of unbalanced data from the KEEL database,and used the model to conduct a classification experiment on these 17 sets of data sets,the results show that the model has a good classification effect on unbalanced data sets;The model is applied in the field of sewage fault diagnosis in later.The experimental results show that the model has a good classification performance for unbalanced sewage data.Finally,in order to further improve the overall classification performance of the model,this paper proposes two integrated optimization models based on Stacking,Multi-Meta Classifier-Stacking(MMC-Stacking)and Weighted Base Classifier Stacking(WBC-Stacking);then the overall framework of the two optimization models are introduced and these two optimization models are applied to the KEEL data set and the sewage data set.The experimental results show that the two optimization models keep the recognition rate of the fault category unchanged but improve the recognition accuracy of most classes,thus the two optimization models improve the overall classification performance.