Planning and Control of Chiller and Power Systems via Collaborative Neurodynamic Optimization
基於協作式神經動力優化的製冷機組和電力系統的規劃與控制
Student thesis: Doctoral Thesis
Author(s)
Related Research Unit(s)
Detail(s)
| Awarding Institution | |
|---|---|
| Supervisors/Advisors |
|
| Award date | 4 Sept 2023 |
Link(s)
| Permanent Link | https://scholars.cityu.edu.hk/en/theses/theses(5a04d775-96b2-4b24-9d63-3fb1f0f8ce26).html |
|---|---|
| Other link(s) |
Abstract
Chiller and power systems are vital facilities in energy systems, accounting for substantial power consumption and costs. In the context of the high demands for reducing energy consumption and carbon emission, it is crucial to increase efficiency in operating systems through planning and control. The complexities of chiller and power systems lead to the nonconvexities of formulated optimization problems for planning and control, bringing challenges in solving processes.
Collaborative neurodynamic optimization (CNO) emerges as a hybrid intelligence paradigm for solving various optimization problems by employing multiple neurodynamic models assisted with a meta-heuristic rule. With its advances in recent years, it would be beneficial and desirable to develop and apply CNO-driven methods in chiller and power systems.
This thesis includes three parts addressing the planning and control of chiller and power systems via CNO. In the first part, global optimization problems are formulated for optimal chiller loading, electrical load dispatch, and chiller plant operation planning, subject to many operational constraints. CNO-driven methods based on multiple projection neural networks are developed to solve them. In the second part, distributed chiller loading is formulated as distributed optimization problems with nonconvex objective functions and discrete feasible regions. To overcome the difficulties caused by the nonconvexities of the formulated problems, CNO-driven methods are developed based on multiple coupled recurrent neural networks or heterogeneous neural networks. In the third part, hybrid model predictive control of chiller systems is formulated as a mixed-integer optimization problem subject to various constraints including thermodynamic and energy-conservation constraints. A CNO-driven method is developed to solve the formulated problem sequentially over a moving time window. The efficacies of these methods in the three parts are demonstrated and substantiated through experimental results based on the specifications and parameters of the systems.
Collaborative neurodynamic optimization (CNO) emerges as a hybrid intelligence paradigm for solving various optimization problems by employing multiple neurodynamic models assisted with a meta-heuristic rule. With its advances in recent years, it would be beneficial and desirable to develop and apply CNO-driven methods in chiller and power systems.
This thesis includes three parts addressing the planning and control of chiller and power systems via CNO. In the first part, global optimization problems are formulated for optimal chiller loading, electrical load dispatch, and chiller plant operation planning, subject to many operational constraints. CNO-driven methods based on multiple projection neural networks are developed to solve them. In the second part, distributed chiller loading is formulated as distributed optimization problems with nonconvex objective functions and discrete feasible regions. To overcome the difficulties caused by the nonconvexities of the formulated problems, CNO-driven methods are developed based on multiple coupled recurrent neural networks or heterogeneous neural networks. In the third part, hybrid model predictive control of chiller systems is formulated as a mixed-integer optimization problem subject to various constraints including thermodynamic and energy-conservation constraints. A CNO-driven method is developed to solve the formulated problem sequentially over a moving time window. The efficacies of these methods in the three parts are demonstrated and substantiated through experimental results based on the specifications and parameters of the systems.
