Fault early warning method based on industrial process operation state trend analysis

Abstract

The invention discloses a fault early warning method based on industrial process operation state trend analysis. The fault early warning method comprises the following steps: step 1, acquiring historical normal process data and process data with faults in an industrial process; 2, monitoring data with faults to obtain values of fault data monitoring statistics; 3, establishing a Gaussian process regression (GPR) model for the fault process data and the observed values thereof; 4, describing the observed value by using qualitative trend analysis to obtain seven basic trend bases, extracting seven basic trends of process data with faults through kernel dictionary learning (KDL), and establishing a trend library of system operation states; and step 5, using a GPR model to predict an observedvalue of the online collection process data, taking the observed value as an input vector of the KDL, and performing classification analysis on different trends to realize early warning of the fault.The method can effectively reflect the fault development trend, and plays an important role in realizing effective fault diagnosis and health management in the industrial process.

Description

technical field [0001] The invention belongs to the field of process industry process fault diagnosis and early warning, and in particular relates to a process industry process fault early warning method based on Gaussian process regression state prediction and kernel dictionary learning trend extraction. Background technique [0002] With the rapid development of the national economy, the scale of industrial production continues to expand, and the modern industrial process is developing in the direction of complexity such as nonlinear, non-Gaussian, unsteady, and multi-modal. Industrial production accidents occur frequently, and accident hazards and losses are huge. , Industrial process safety is directly related to national economic development and the safety of people's lives and property. Process safety, product quality, energy saving, emission reduction and efficiency enhancement have gradually become the core goals of modern industry. Fault trend prediction is an impor...

AI Technical Summary

Problems solved by technology

At present, the main methods of industrial process operation state prediction include autoregressive moving average method, support vector regression and neural network methods; it is difficult to determine the parameters of industrial process time series data by autoregressive moving average method, resulting in different time series process data. Insufficient prediction robustness and other issues; support vector regression is very difficult for some complex nonlinear industrial process data prediction; neural network methods require a large number of training samples and are prone to problems such as over-fitting, making it difficult to make effective predictions

Images