Nonlinear Chemical Process Fault Detection Method

Abstract

The invention relates to a non-linear chemical process fault detection method. The specific steps are as follows: in the offline modeling stage, first normalize the training data, then construct M random slow feature analysis models, and obtain the corresponding dominant Control limits of subspace statistics and residual space statistics; in the online detection stage, the test data is collected and normalized, and then mapped to M random slow feature analysis models to obtain M groups of dominant subspace statistics and Residual spatial statistics, integrated monitoring statistics and statistical BIC obtained through the weighted probability index fusion mechanism Q , respectively compared with 1‑α to judge whether a fault occurs, where α is the confidence level. The invention utilizes the random Fourier mapping to dig deep into the data characteristics of the nonlinear process, can more efficiently establish a nonlinear slow feature analysis model, and improves the fault detection effect.

Description

technical field [0001] The invention belongs to the technical field of industrial process detection and relates to a nonlinear chemical process fault detection technology, in particular to a nonlinear chemical process fault detection method based on random slow feature analysis. Background technique [0002] In the process of modern industrial production, real-time fault detection technology has become an important support to ensure safe production and improve product quality. Due to various failures such as instrument failure, valve stickiness, material leakage, etc. may occur during the production process, these failures may cause measurement deviation and product quality reduction, and cause dangerous safety accidents, resulting in equipment damage and casualties. directly affect the normal operation of the enterprise. Therefore, researchers and engineers have been working on advanced process monitoring and fault detection techniques. Among them, the data-driven method ...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the above-mentioned problems of high computational complexity and low fault detection sensitivity in the prior art, the present invention provides a non-linear chemical process fault detection method based on stochastic slow feature analysis (SFA for short), which can improve the calculation cost of traditional kernel matrix. When the problem, improve the fault detection sensitivity

Images