Three-layer deep learning network random trees for fault detection in chemical production process

Ming Lu,Zhen Gao,Ying Zou,Zuguo Chen,Pei Li

2024-08-10

Abstract:With the development of technology, the chemical production process is becoming increasingly complex and large-scale, making fault detection particularly important. However, current detective methods struggle to address the complexities of large-scale production processes. In this paper, we integrate the strengths of deep learning and machine learning technologies, combining the advantages of bidirectional long and short-term memory neural networks, fully connected neural networks, and the extra trees algorithm to propose a novel fault detection model named three-layer deep learning network random trees (TDLN-trees). First, the deep learning component extracts temporal features from industrial data, combining and transforming them into a higher-level data representation. Second, the machine learning component processes and classifies the features extracted in the first step. An experimental analysis based on the Tennessee Eastman process verifies the superiority of the proposed method.

Machine Learning

What problem does this paper attempt to address?

The paper aims to address the issue of fault detection in chemical production processes. With the advancement of technology, chemical production processes have become increasingly complex and large-scale, making fault detection particularly important. However, current detection methods struggle to cope with the complexity of large-scale production processes. To solve this problem, the authors propose a new fault detection model called Three-Layer Deep Learning Network Random Trees (TDLN-trees). TDLN-trees combine the advantages of deep learning and machine learning technologies, utilizing Bidirectional Long Short-Term Memory (BLSTM), Fully Connected Neural Network (FCNN), and Extra Trees Algorithm (ET) to extract and process temporal features in industrial data, and classify these features to achieve fault detection. The workflow of the TDLN-trees model proposed in the paper is as follows: 1. **Data Extraction and Preprocessing**: First, the sliding window method is used to extract feature matrices with temporal correlations, then the data is normalized, and the labels are one-hot encoded. 2. **Three-Layer Deep Learning Components**: - First Layer: Use the BLSTM layer to capture the temporal dynamics of the data. - Second Layer: Use the LSTM layer to further enhance the model's ability to capture short-term dependencies. - Third Layer: Use the FCNN layer to combine and transform temporal features to enhance the model's ability to interpret high-level data. 3. **Machine Learning Component**: The output of the FCNN layer is passed to the ET, which uses the Gini index to select split points, thereby mapping process variable features to corresponding fault types. To validate the effectiveness of the proposed method, the authors conducted experimental analysis on the Tennessee Eastman Process (TEP) dataset. The experimental results show that TDLN-trees achieved a fault detection rate of 98.46%, outperforming other comparison methods, demonstrating the advantages of TDLN-trees in handling large-scale, high-dimensional, and nonlinear data. In summary, the goal of this paper is to develop a new method capable of effectively handling fault detection in chemical production processes. By integrating the technological advantages of deep learning and machine learning, the TDLN-trees model is proposed and its superior performance is validated through experiments.

Three-layer deep learning network random trees for fault detection in chemical production process

Three‐layer deep learning network random trees for fault detection in chemical production process

Application of Neural Networks to Chemical Fault Diagnosis

Deep Learning With Tensor Factorization Layers For Sequential Fault Diagnosis And Industrial Process Monitoring

Multichannel Dynamic Graph Convolutional Network-Based Fault Diagnosis and Its Application in Blast Furnace Ironmaking Process

A Novel Fault Detection and Identification Method for Complex Chemical Processes Based on OSCAE and CNN

Intelligent Fault Diagnosis for Chemical Processes Using Deep Learning Multimodel Fusion

Deep Convolutional Neural Network Model Based Chemical Process Fault Diagnosis

Fault Diagnosis of the Dynamic Chemical Process Based on the Optimized CNN-LSTM Network

A deep belief network based fault diagnosis model for complex chemical processes

Integrating feature optimization using a dynamic convolutional neural network for chemical process supervised fault classification

Deep ensemble forests for industrial fault classification

An optimized long short-term memory network based fault diagnosis model for chemical processes

High-fidelity Positive-Unlabeled Deep Learning for Semi-Supervised Fault Detection of Chemical Processes

A Novel Label-Aware Global Graph Construction Method and Spiking-Coded Graph Neural Network for Intelligent Process Fault Diagnosis

Dynamic Ensemble Selection Based Improved Random Forests for Fault Classification in Industrial Processes

Fault Detection and Diagnosis Based on Transfer Learning for Multimode Chemical Processes

A Novel Transformer-Based Model with Large Kernel Temporal Convolution for Chemical Process Fault Detection

An Order-Invariant and Interpretable Dilated Convolution Neural Network for Chemical Process Fault Detection and Diagnosis

Hierarchical Deep Recurrent Neural Network based Method for Fault Detection and Diagnosis

Fault Diagnosis of Intelligent Production Line Based on Digital Twin and Improved Random Forest