An online adaptive fault monitoring and diagnosis method for process industry process

Abstract

The invention discloses an on-line self-adaptive working condition monitoring and fault diagnosis method of a process industrial process, which belongs to the technical field of fault monitoring and diagnosis of complex industrial processes. The present invention first analyzes historical observation data under normal working conditions, and establishes an industrial process fault monitoring model based on sparse principal component analysis by introducing an elastic regression network combined with Lasso and Ridge constraints to obtain industrial process fault monitoring statistics. Process control limit; in the online monitoring of industrial process faults, the process of decomposing the covariance matrix of online monitoring data by using the rank-1 matrix correction algorithm, and recursively updating the load matrix of the sparse monitoring model to obtain process fault monitoring statistics suitable for the working conditions control limits to realize self-adaptive detection of process faults in the process industry; finally, for the detected faults, the contribution graph method is used to obtain the specific causes of the faults. The invention can self-adaptively monitor the faults of process industrial processes with complex and changeable working conditions for a long time, and has the advantages of low computational complexity, high precision, low false alarm rate and the like.

Description

technical field [0001] The invention relates to the field of industrial process automation monitoring, in particular to a method and technology for fault monitoring and diagnosis of process industrial processes. Background technique [0002] Process industry, also known as process industry, refers to the production process through physical changes and chemical changes. The process industry mainly includes basic raw material industries such as petroleum, chemical industry, steel, non-ferrous metals, and building materials. It is the pillar and basic industry of the national economy and an important supporting force for my country's sustained economic growth. [0003] Process safety, product quality, energy saving, emission reduction and efficiency enhancement are the core goals of the modern process industry. A good industrial process operating condition is the key to stabilizing production indicators, ensuring product quality, and realizing stable and optimized operation of ...

AI Technical Summary

Problems solved by technology

[0006] However, the traditional multivariate statistical analysis method usually assumes that the process variables are static, that is, the linear relationship between the variables does not change with the operation of the process. However, in most actual industrial production processes, the raw material composition of process production is complex Various factors such as variability, human disturbance in process operation, wear and tear of production equipment, sensor drift, and environmental differences in physical and chemical reactions in the process will cause changes in the state of industrial processes.

In this case, the normal working condition monitoring model based on static historical data is often unable to adapt to the dynamic changes of complex industrial processes, especially for process monitoring and fault diagnosis of process The model is extremely prone to problems such as false positives, missed negatives, and insufficient response

Images