Quality Monitoring and Root Cause Diagnosis for Industrial Processes Based on Lasso-SAE-CCA

Jie Dong,Liang Ma,Kaixiang Peng,Ruiqi Sun,Zhijie Shi

doi:10.1109/access.2019.2926067

Jie Dong, Liang Ma + Show 3 more

Open Access

https://doi.org/10.1109/access.2019.2926067

Copy DOI

Journal: IEEE Access	Publication Date: Jan 1, 2019
Citations: 11	License type: CC BY 4.0

Affiliation: University of Science and Technology Beijing

Abstract

This paper is concerned with theoretical and practices approach for overall quality-related fault detection and identification in process industries. Fault detection and fault tracing can help engineers to take corrective actions and recover the process operations. A novel diagnostic method is proposed based on stacked automatic encoder-canonical correlation analysis (SAE-CCA) and least absolute shrinkage selection operator (Lasso). First, a quality monitoring scheme based on SAE-CCA is proposed, which establishes a relationship model among quality and characteristic variables to detect faults. Then, Lasso is used for locating the root causes, according to the process state and fault information. Finally, the experiments are conducted with typical industry process data, i.e., a hot strip mill process (HSMP), in order to demonstrate the effectiveness of the whole diagnosis method.

Highlights

With the development of manufacturers and increasement in quality demand, industrial systems tend to be more integrated and complicated
This study focuses on extracting the effective data features and maximizes the correlation among characteristic and quality variables based on stacked auto-encoder (SAE)-Canonical Correlation Analysis (CCA) joint driven algorithm
In this study, a novel quality-related diagnostic method based on least absolute shrinkage selection operator (Lasso)-SAE-CCA is proposed

Summary

INTRODUCTION

With the development of manufacturers and increasement in quality demand, industrial systems tend to be more integrated and complicated. A data-based method can construct a causal topology model by mining association information among process variables. It is intensely popular in the industrial and academic fields because of little need for deep knowledge and mechanism models. J. Dong et al.: Quality Monitoring and Root Cause Diagnosis for Industrial Processes Based on Lasso-SAE-CCA and predicted by obtaining an optimal combination of process variables in the MLR [9]. The fault variables are selected by adjusting the penalty coefficient This method is different from the contribution plot which rely on monitoring index completely [21].

PRINCIPLE AND PROBLEM FORMULATION

BASIC IDEA OF CCA ALGORITHM

LASSO VARIABLE SELECTION ALGORITHM

LASSO- BASED FAULT DIAGNOSIS

SIMULATION STUDY

Findings

CONCLUSION