Air Separation Process Monitoring And Fault Diagnosis Based On Multivariate Statistical Process Control

With the scientific and technological progress, air separation industry has been booming. Meanwhile, air separation plant is becoming more and more large-scale and complex. The requirement for energy saving and safety has been a key point of the air separation products. Fault diagnosis is critical to reducing accidents in production, improving production performance, saving economic losses, and enhancing the competitiveness of enterprises. It is an important component of production management in the modern enterprise, including the air separation enterprises. Air separation process monitoring and fault diagnoses system, which can effectively use the data information of the process, is capable of timely detecting the faults, fast locating the faults, profoundly analyzing the cause of the faults, and thus improving stability and efficiency of the air separation production.Multivariate Statistical Process Control (MSPC) is a method for monitoring, analyzing and controlling the process operating performance. By the statistical analysis of the data, MSPC can closely monitor the production process, constantly test the changes of the process with the information of the faults, timely identify and resolve the faults, and then analyze the potential trend of the faults so as to make a prediction of the faults. Therefore, fault diagnosis technology based on MSPC has important practical value for the industry. It has become a research hotspot.In this dissertation, we studied several important aspects of MSPC and fault diagnosis technology. Combining the characteristics of the air separation process, we created an air separation monitoring and fault diagnosis system, which achieves good results in practical applications. The main research work includes as following:1. Fault diagnosis technology and MSPC are generally introduced, including an overview of the relevant basic concepts, methods and the developments. We also introduce the status quo of the air separation process fault diagnosis, in which we point out the significance and methods of the research in this dissertation. 2. By studying the air separation process and the application requirements of MSPC, we make a necessity and feasibility analysis on the method, and then raise the overall design the method.3. Concepts and method of the Principal Component Analysis (PCA) and multivariate statistical control charts are introduced. And then an air separation monitoring and fault diagnosis model based on PCA is established. In the modeling process, we focus on the data process in which the modeling data is generated from the industrial raw data. The diagnostic results of the model are analyzed and the improvement programs are proposed.4. Dynamic Principal Component Analysis (DPCA) is introduced. Most of all, the methods of determining the time-lag is described, and we propose a new method to determine the time-lag by the Model Evaluation. After studying the dynamic characteristics of the air separation process, we establish an air separation process monitoring and fault diagnosis model based on DCPA. The diagnostic results of the model are analyzed, which verify the accuracy of the model. Meanwhile, we discuss the further explore scheme, which is the Optimal Time-lag Configuration DPCA method.5. The design and operating results of the air separation monitoring and fault diagnosis system are introduced. We highlight the interface of the fault diagnosis system, which has a rich amount of information and vivid dynamic display effect. The practical operation results illustrate the validity of our study.In the end, the dissertation is concluded with a summary and discussions of the prospective research on open problems.