Fault Diagnosis Method For Fused Magnesium Oxide Industrial Process Based On Data

With the development of science and technology,industrial systems in modern large-scale production have become larger and more complex.In the event of failure of these systems,they will cause huge property damage and casualties.Therefore,improving the reliability and maintainability of complex dynamic systems has become the primary issue in the current enterprise development.The data-driven fault diagnosis technology is based on the massive data collected in the actual industrial production process,using various data processing methods,statistical modeling methods,etc.to establish an appropriate mathematical model for industrial processes within a certain range,thereby achieving the purpose of detecting and diagnosing whether a malfunction has occurred in an industrial process.The research of this technology not only has many theoretical significance,but also has a wide application background,which is an important development direction of future fault diagnosis technology.In this paper,the fault diagnosis research is conducted with the background of fused magnesia production process.Based on the modeling and analysis of the image data collected in the production process,a fault diagnosis method based on the image data is proposed.Aiming at the multi-source heterogeneous data that can be collected in the production process,a method for fault diagnosis of heterogeneous data based on feature regression is proposed,which solves the problems of collaborative modeling of heterogeneous data and fault diagnosis.The research work done in this article is as follows:(1)In order to solve the problems found in the fault diagnosis based on image data,a single-view data industrial fault detection method based on non-negative matrix factorization is proposed.The image data under normal working conditions collected in actual production is modeled and applied to industrial fault detection.Due to the limitation of single-view data,an industrial fault detection method based on shared subspace multi-view non-negative matrix factorization is proposed,and a process monitoring model for fused magnesia production process is established.The experimental simulation shows that the algorithm can effectively identify the operating state of the fused magnesia melting process.(2)Since data from different data sources can be collected during the production process,the data of a single data source can only detect a single abnormal condition during fault detection,and cannot comprehensively and accurately detect the entire production process.Therefore,in order to solve this problem,this paper presents a method for fault diagnosis of heterogeneous data based on feature regression,and collaboratively models the collected data from different data sources,and applies it to the fused magnesium oxide production process.The experimental results show that the method can quickly and effectively identify the operating state of the fused magnesium oxide smelting process,and can make an accurate judgment on the type of abnormal working conditions.(3)The fault detection method proposed in this paper is applied to the fused magnesium oxide production process,and the abnormal operating conditions are controlled according to the previous research on the arc model and the actual production process requirements,so that the smelting process is more stable and efficient.