Research On Complex Process Fault Diagnosis Method Based On Deep Learning

With the rapid development of science and technology and the increasing complexity of industrial systems,the efficient and safe operation of the support system has become the key to improving product quality and improving production efficiency.Complex processes are characterized by complex processes of multiple stable operating conditions due to external environments,inherent properties of industrial systems,different operating conditions,and different production requirements.To ensure the safe and efficient operation of multiple complex processes,it is especially necessary to establish an accurate and effective fault diagnosis model.Therefore,effective fault diagnosis of multi-modal complex industrial processes is one of the research hotspots in the field of fault diagnosis.In this thesis,the fault diagnosis method is studied by using the simulation model of the Tennessee Eastman Process(TEP).For the complex process,the focus of this thesis is to design and build an experimental platform based on deep learning fault diagnosis method,and propose two kinds of multi-modal process fault diagnosis methods based on deep learning,which are implemented on the experimental platform.In this paper,the fault diagnosis method is studied by using the simulation model of Tenessee Eastman Process(TEP).For complex processes,the focus of this paper is to design and build an experimental platform based on deep learning fault diagnosis method,and propose two deep learning based multi-modal process fault diagnosis methods,which are implemented on the experimental platform.Since the data generated in the actual industrial process is often contalinated by noise,in order to accurately and stably extract the fault characteristics,this paper proposes the Variational Mode Decomposition(VMD)and Stacked Auto-encoder(SAE),VMD and Convolutional Neural Network(CNN)combined two types of fault diagnosis methods,and used for TE process fault diagnosis.In this paper,deep learning is applied to the fault diagnosis scheme.Through the proposed method,the deep architecture of fault data is learned to minimize information loss and prevent noise interference,and the softmax classifier is used for fault diagnosis,thereby improving fault detection.The accuracy of the classification and effectively solve the failures that cannot be detected by the traditional fault diagnosis technology,so as to achieve the purpose of improving the accuracy of the fault diagnosis.The results show that the proposed method not only improves the separability between fault and normal process,but also shows the accuracy and validity of fault classification for Tennessee Eastman process data.The accuracy of fault diagnosis of the proposed two methods is higher than that of machine learning and existing deep learning methods.The fault diagnosis method based on deep learning is faster than the existing fault diagnosis method based on deep learning.In the proposed method,VMD-CNN is superior to VMD-SAE in the accuracy and speed of fault diagnosis.