Multi-Rate Fault Detection Using Dynamic and Static Latent Variables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing industrial process monitoring models fail to effectively handle multi-rate characteristics of process data, leading to compromised fault detection in complex industrial processes due to the neglect of both dynamic and static characteristics.
Innovation Solution
A multi-rate industrial process fault detection method based on a Total Auto-regressive Dynamic Latent Variable Model (TMrARDLV) that utilizes multi-rate data samples, considering both dynamic and static characteristics through a structured model with sampling coefficients, and employs EM algorithm, Kalman filtering, and Bayesian methods for accurate latent variable estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional static process monitoring models (PCA, PLS) are used, then the model structure is simple, but fault detection performance deteriorates when dealing with high-dimensional time-series correlated process data
Solution Approach 1:
The patent transitions from static monitoring models to dynamic monitoring models that capture temporal correlations in process data. The dynamic latent variable models incorporate time-dependent relationships through state space representations and autoregressive structures, enabling the system to adapt to changing process conditions while maintaining detection accuracy.
Solution Approach 2:
The patent changes the fundamental parameters of the monitoring model by introducing dynamic latent variables that evolve over time. The model parameters include state transition matrices, observation matrices, and covariance structures that capture the temporal dynamics of multi-rate process data, fundamentally improving fault detection capability compared to static models.
2Reliability
If dynamic process monitoring models (DPCA, CVA, LGSSM) are used, then temporal correlations are described, but multi-rate characteristics of process data are not considered
Solution Approach 1:
The patent segments the process monitoring model into multiple independent dynamic latent variable models, each operating at a specific sampling rate. This segmentation allows each sub-model to handle data at its native rate while maintaining temporal correlations, and the results are integrated through a unified fault detection framework.
Solution Approach 2:
The patent creates a universal monitoring framework that can handle multiple sampling rates simultaneously. The multi-rate dynamic latent variable model serves multiple functions: it processes fast-sampled process variables, intermediate variables, and slow-sampled quality variables within a single unified structure, making the system adaptable to diverse multi-rate process configurations.
3Adaptability or versatility
If Multi-rate Dynamic Latent Variable Model is used, then multiple sampling information is utilized, but static characteristics are neglected leading to compromised fault detection in strongly coupled complex processes
Solution Approach 1:
The patent merges dynamic and static characteristics into a unified Total Auto-regressive Dynamic Latent Variable (TADLV) model. The model combines the dynamic latent variables that capture temporal correlations with static latent variables that represent steady-state relationships, creating a comprehensive monitoring approach that leverages both dynamic and static process characteristics for improved fault detection.
Solution Approach 2:
The patent creates a composite monitoring model structure that integrates multiple components: dynamic latent variable models for temporal dynamics, static latent variable models for steady-state relationships, and autoregressive structures for additional temporal dependence. This composite structure effectively handles strongly coupled complex processes by combining complementary modeling approaches.
Data Source
AI summary
The present invention discloses a multi-rate process fault detection method based on a Total Auto-regressive Dynamic Latent Variable Model. The multi-rate data samples of the process are collected online. The method utilizes a Total Multi-rate Auto-regressive Dynamic Latent Variable Model (TMrARDLV) to obtain the dynamic T2 statistics of the current moment test samples, static T2 statistics for each sampling rate, and SPE statistics. These statistics are then compared with the pre-established detection control limits to determine the online detection results of the process. This method fully utilizes comprehensive multi-rate data information from the process. It also considers the dynamic and static characteristics of the data separately by employing Kalman filtering and Bayesian methods. Moreover, it achieves accurate estimation of dynamic and static latent variables. The dynamic and static latent variables obtained through dimensionality reduction respond to faults in different data subspaces. This method enhances the accuracy and applicability of fault detection.