Tesi etd-09062019-112449
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
DETTORI, STEFANO
Indirizzo email
ste.dettori@gmail.com
URN
etd-09062019-112449
Titolo
Advanced Modelling and Control Methodologies for Energy Optimization in Industrial Contexts
Settore scientifico disciplinare
ING-IND/13
Corso di studi
INGEGNERIA - Ph.D. Programme in Emerging Digital Technologies (EDT)
Commissione
Presidente Prof. FRISOLI, ANTONIO
Membro Prof. VALENTINI, RENZO
Membro Prof. ALLOTTA, BENEDETTO
Membro Dott.ssa COLLA, VALENTINA
Membro Prof. VALENTINI, RENZO
Membro Prof. ALLOTTA, BENEDETTO
Membro Dott.ssa COLLA, VALENTINA
Parole chiave
- Artificial Intelligence
- Deep Echo State Networks
- Hierarchical Control Systems
- Integrated Steelworks
- Model Predictive Control
- Process Off-Gas Networks
- Rotor Stress control
- Steam Turbine Startup Controller
- Steam Turbines
Data inizio appello
18/06/2020;
Disponibilità
parziale
Riassunto analitico
Nowadays, large manufacturing industries, energy intensive enterprises and energy producers are going through a period of unprecedented challenges. On one hand, the market of raw materials and energy is much more dynamic than in the past, with the consequence of an oscillation in ever shorter periods of their prices. On the other hand, the production of manufactured goods in terms of quantity and quality has assumed, for more than a decade (with the 2007/8 market crisis), a discontinuous behavior. While the processes were previously designed and consequently controlled so that their operating points were as constant as possible and around maximum energy efficiency, now their scheduling often shows huge variations between the production peaks and the relative lows, depending on market fluctuations.
In this context, an increasingly widespread awareness of environmental issues and of safeguarding the relative balances is being developed and a consciousness was born in the scientific world starting from the efforts of the Club of Rome (1968). These factors led to revising the mid-80s philosophy of the current linear economic system, which favors the disposal of waste and therefore uses only virgin materials, and converting it into a more modern concept of circular economy. The Circular Economy approach aims at extending the life of the products, reconditioning them and reducing waste production, also in terms of re-use and recovery of all energy sources related to the by-products of the processes and their valorization.
This sensitivity to environmental issues has been developed relatively recently by sectors of process industry, with the consequent search for new solutions for production processes or improvement of flexibility of the old ones. Such solutions require rethinking the philosophy of the dated monitoring, supervision and control systems, based on standard techniques that do not allow achieving sufficient performances when the process setpoints frequently change, or that do not even give sufficient guarantees in terms of safety, with the consequent degradation of production rates. These problems can be effectively addressed through modern artificial intelligence techniques and advanced control systems, which allow achieving the objectives of flexibility, safety and productivity and at the same time minimize the environmental impact taking into consideration also energy optimization, recovery and re-use of by-products and the consequent eco-sustainability.
This PhD thesis focuses and deepens two points. The first one is to develop modeling and control techniques applied to complex industrial systems with objectives linked to the energy optimization of their processes. The second aspect is related to the technological transfer of the knowledge acquired by the research towards the industrial world also through the development of algorithms that can be implemented in control systems, which are typically used in industry, by taking into account their requirements in terms of computational performance.
In particular, this thesis explores topics of advanced modeling based on hybrid approaches of classical, physical based and artificial intelligence techniques for the monitoring of process variables, and their use within predictive Model control philosophies.
The approach followed is deepened in the thesis also through the presentation of case studies that allow highlight some of the problems to be faced in the control complex industrial systems, in particular focusing on the relative customization of the algorithms that must be tailored for the specific calculation platform and the specific process. In particular, solutions are presented for intrinsically non-linear systems or hybrid processes (mixed logical dynamical systems), and hierarchical and distributed control architectures.
In this context, an increasingly widespread awareness of environmental issues and of safeguarding the relative balances is being developed and a consciousness was born in the scientific world starting from the efforts of the Club of Rome (1968). These factors led to revising the mid-80s philosophy of the current linear economic system, which favors the disposal of waste and therefore uses only virgin materials, and converting it into a more modern concept of circular economy. The Circular Economy approach aims at extending the life of the products, reconditioning them and reducing waste production, also in terms of re-use and recovery of all energy sources related to the by-products of the processes and their valorization.
This sensitivity to environmental issues has been developed relatively recently by sectors of process industry, with the consequent search for new solutions for production processes or improvement of flexibility of the old ones. Such solutions require rethinking the philosophy of the dated monitoring, supervision and control systems, based on standard techniques that do not allow achieving sufficient performances when the process setpoints frequently change, or that do not even give sufficient guarantees in terms of safety, with the consequent degradation of production rates. These problems can be effectively addressed through modern artificial intelligence techniques and advanced control systems, which allow achieving the objectives of flexibility, safety and productivity and at the same time minimize the environmental impact taking into consideration also energy optimization, recovery and re-use of by-products and the consequent eco-sustainability.
This PhD thesis focuses and deepens two points. The first one is to develop modeling and control techniques applied to complex industrial systems with objectives linked to the energy optimization of their processes. The second aspect is related to the technological transfer of the knowledge acquired by the research towards the industrial world also through the development of algorithms that can be implemented in control systems, which are typically used in industry, by taking into account their requirements in terms of computational performance.
In particular, this thesis explores topics of advanced modeling based on hybrid approaches of classical, physical based and artificial intelligence techniques for the monitoring of process variables, and their use within predictive Model control philosophies.
The approach followed is deepened in the thesis also through the presentation of case studies that allow highlight some of the problems to be faced in the control complex industrial systems, in particular focusing on the relative customization of the algorithms that must be tailored for the specific calculation platform and the specific process. In particular, solutions are presented for intrinsically non-linear systems or hybrid processes (mixed logical dynamical systems), and hierarchical and distributed control architectures.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|