A Hardware-in-Loop Digital Twin Approach for Intelligent Optimization of Municipal Solid Waste Incineration (eBook)

AI and Its Application to Complex Industrial Processes

Jiang Tang, Wen Yu, Junfei Qiao (Autoren)

eBook Download: EPUB

2025
940 Seiten
Wiley-IEEE Press (Verlag)
978-1-394-35402-3 (ISBN)

Infos

An expert discussion of intelligent optimization control in complex industrial processes

In A Hardware-in-Loop Digital Twin Approach for Intelligent Optimization of Municipal Solid Waste Incineration: AI and Its Application to Complex Industrial Processes, a team of distinguished researchers delivers an innovative new approach to integrating virtual mechanism data generated through coupled numerical simulation and orthogonal experimental design with real historical data. The book explains how to create a heterogenous ensemble prediction model for carbon monoxide emissions in municipal solid waste incineration (MSWI) processes.

The authors focus on intelligent optimization control of MSWI processes based on hardware-in-loop DT platforms. They demonstrate AI-driven modeling, control, optimization algorithms in real-world applications, including virtual-real data hybrid-driven deep modeling and intelligent optimal controls based on multiple objectives.

Additional topics include:

• A thorough introduction to numerical simulation modeling of whole industrial processes
• Comprehensive explorations of the design, implementation, and validation of hardware-in-loop digital twin platforms
• Practical discussions of AI-driven modeling, control, and optimization
• Fulsome descriptions of the skills required to address challenges posed by complex industrial processes

Perfect for environmental engineers and researchers, A Hardware-in-Loop Digital Twin Approach for Intelligent Optimization of Municipal Solid Waste Incineration will also benefit MSWI plant operators and managers, as well as AI and machine learning researchers and developers of environmental monitoring and control systems.

Jian Tang, PhD, is a Professor and Researcher with the Department of Artificial Intelligence and Automation in the Faculty of Information Technology at the Beijing University of Technology.

Wen Yu, PhD, is a Professor and Head of Department of the Departamento de Control Automatico at CINVESTAV-IPN (National Polytechnic Institute) in Mexico City, Mexico.

Junfei Qiao, PhD, is a Professor with the Beijing University of Technology and Director of Beijing Laboratory of Smart Environmental Protection in Beijing, China.

Textauszug

List of Figures

Figure 1.1

Process flow of a grate‐type MSWI plant in Beijing

Figure 1.2

MSW components ratios of different countries/regions

Figure 1.3

Schematic diagram of manual control mode in the MSWI process in China

Figure 1.4

Schematic diagram of operation optimization process in complex process industries

Figure 1.5

Requirements and relationships in academic research and industrial applications

Figure 1.6

Relationship between the incineration mechanism, the actual MSWI process, human brain cognitive theory, numerical simulation challenges, and DT model construction

Figure 1.7

Structure of “Real–Real” simulation platform

Figure 1.8

Structure of “Real–Virtual” simulation platform

Figure 1.9

Structure of “Virtual–Real” simulation platform

Figure 1.10

Structure of “Virtual–Virtual” simulation platform

Figure 1.11

The book's structure

Figure 2.1

Process flows of MSWI plants with a daily processing capacity of 800 tons

Figure 2.2

Internal structure and zoning diagram of mechanical grate furnace

Figure 2.3

Diagram of flue gas cleaning process

Figure 2.4

Solid‐phase combustion zone and gas‐phase combustion zone for nitrogen element products

Figure 2.5

Schematic diagram of NxOy generation in high‐temperature combustion area

Figure 2.6

Numerical simulation and modeling analysis framework

Figure 2.7

Simulation modeling strategy for MSWI whole process

Figure 2.8

Multi‐software‐coupled whole‐process numerical simulation strategy under benchmark conditions

Figure 2.9

Incinerator simplified structure (left side), its 2D model (middle), and mesh division (right side)

Figure 2.10

Non‐grate solid‐phase combustion simulated by Aspen Plus

Figure 2.11

Combustion results of solid MSW on the grate. (a) Rate; (b) Mass fraction

Figure 2.12

Combustion results of gas‐phase combustion under benchmark conditions: (a) temperature, (b) O2 mass fraction, and (c) CO2 mass fraction

Figure 2.13

Temperature distribution in the incinerator under typical. (a–h) Case 1–Case 8

Figure 2.14

O2 distribution in the incinerator under typical. (a–h) Case 1–Case 8

Figure 2.15

CO2 distribution in the incinerator under typical. (a–h) Case 1–Case 8

Figure 2.16

Probability density of temperature in the incinerator under typical. (a–h) Case 1–Case 8

Figure 2.17

Exhaust emission results obtained from the simulation: (a) CO, (b) CO2, (c) O2, (d) SO2, and (e) NOx

Figure 2.18

Single factor analysis curve in terms of feed rate based on MIMO‐LRDT mechanism model. (a) CO concentration; (b) CO2 concentration; (c) O2 concentration; (d) SO2 concentration; (e) NOx concentration

Figure 2.19

Single factor analysis curve in terms of primary air temperature based on MIMO‐LRDT mechanism model. (a) CO concentration; (b) CO2 concentration; (c) O2 concentration; (d) SO2 concentration; (e) NOx concentration

Figure 2.20

Dual‐factor analysis curve based on MIMO‐LRDT mechanism model. (a) Grate speed vs Feed rate; (b) Grate speed vs Primary air temperature

Figure 3.1

Strategy diagram of the proposed virtual data and real data hybrid‐driven modeling approach

Figure 3.2

Aspen Plus model diagram

Figure 3.3

Structure diagram of MISO LRDT model

Figure 3.4

LSTM structure diagram

Figure 3.5

Impact of three inputs on CO under multi‐operating conditions

Figure 3.6

Prediction curves of different models based on virtual mechanism data

Figure 3.7

Prediction curves of different models based on real data

Figure 3.8

Prediction curves of different models for offline training verification phase

Figure 3.9

Prediction curves of the offline training verification phase

Figure 3.10

Prediction curves of different models for the online testing verification phase

Figure 3.11

Prediction curves of online testing verification phase

Figure 3.12

Relationship between the hyperparameter and R2 indicator

Figure 4.1

DXN generation mode during MSW combustion process

Figure 4.2

DXN generation mode after MSW combustion process

Figure 4.3

Schematic diagram of the generation mechanism and temperature range of DXN

Figure 4.4

SEN modeling strategy based on Bayesian inference and binary tree

Figure 4.5

Schematic diagram of BT candidate submodels construction

Figure 4.6

Prediction curves of the candidate submodels for the benchmark datasets

Figure 4.7

Posterior information of the candidate submodels for the benchmark datasets

Figure 4.8

Posterior information of the selected ensemble submodels for the benchmark datasets

Figure 4.9

Fitting curves of the SEN model for the benchmark datasets

Figure 4.10

Posterior information of the SEN model for the DXN dataset

Figure 4.11

Posterior information of the selected ensemble submodels for the DXN dataset

Figure 4.12

Fitting curves of the DXN dataset

Figure 4.13

Hyperparameter sensitivity analysis curves of the BBTSEN model

Figure 5.1

Semi‐supervised RF optimization strategy for DXN emission soft sensing

Figure 5.2

Schematic of the parameter coding design for the semi‐supervised RF optimization

Figure 5.3

Particle decoding schematic for the semi‐supervised RF optimization strategy

Figure 5.4

Modeling results after CCS dataset optimization for the semi‐supervised RF optimization strategy

Figure 5.5

Prediction curve of the testing set on the CCS data for the semi‐supervised RF optimization strategy

Figure 5.6

Modeling results of the DXN dataset for the semi‐supervised RF optimization strategy

Figure 5.7

Prediction curve of the testing set on the DXN data for the semi‐supervised RF optimization strategy

Figure 5.8

Relationship between the hyperparameters and RMSE in the CCS original dataset for the semi‐supervised RF optimization strategy

Figure 5.9

Relationship between the hyperparameters and RMSE in the CCS mixed dataset for the semi‐supervised RF optimization strategy

Figure 5.10

Relationship between the hyperparameters and RMSE in the DXN original dataset for the...

Produkt-Details

• Erscheint lt. Verlag: 5.11.2025
• Sprache: englisch
• Themenwelt: Technik ► Bauwesen
• Themenwelt: Technik ► Elektrotechnik / Energietechnik
• Schlagworte: Artificial Intelligence • Digital Twin (DT) • ensemble learning • hybrid data-driven • Industrial Control • industrial modeling • Industrial Optimization • municipal solid waste incineration (MSWI) • numerical simulation • Semi-Supervised Learning
• ISBN-10: 1-394-35402-9 / 1394354029
• ISBN-13: 978-1-394-35402-3 / 9781394354023