|Title:||Agent-based distributed real-time optimal control for building HVAC systems implemented on IoT platforms|
|Advisors:||Wang, Shengwei (BEEE)|
|Subject:||Buildings -- Energy conservation|
Buildings -- Environmental engineering
Internet of things
Hong Kong Polytechnic University -- Dissertations
|Department:||Department of Building Environment and Energy Engineering|
|Pages:||xi, vii, 121 pages : color illustrations|
|Abstract:||HVAC systems for buildings consume huge amounts of energy, many researchers have made serious efforts to develop real-time optimal control or supervisory control strategies to enhance system energy efficiency. However, the centralized form of the existing optimal control strategies results in several drawbacks, including the lack of generality and flexibility and the dependency on central computation stations. On the other hand, with the rapid development of related technologies, Internet of Things (IoT) has been attracting increasing attention in various industries, including its applications in building automation systems (BASs). The implementation of a large amount of IoT devices enables new applications and improves the existing ones but also brings new challenges in network traffic and system reliability. Distributed optimal control strategies offer effective means to avoid the drawbacks of centralized optimal control strategies and have the potential to be implemented on the field level of BASs to reduce the network traffic and increase control reliability. However, existing studies on distributed optimal control of HVAC systems rarely consider the needs and constraints for practical applications and the deployment of control strategies on the physical building automation platforms. The following problems and challenges when developing the distributed optimal control strategies for real applications are not well addressed: i. the computation loads of the optimization tasks need to be distributed appropriately to allow them to be handled by local devices; ii. convergence of the optimization needs to be achieved within the required optimal control interval; iii. the impacts of information delays on control performance need to be investigated; iv. distributed optimal control strategies developed should have the capability of reducing the impacts of information delays. This PhD study, therefore, aims to develop agent-based distributed real-time optimal control strategies for the building HVAC systems concerning the deployment on the local controllers of current LAN-based field control networks and the smart sensors of future IoT-enabled field control networks.|
To address the first challenge, agent-based distributed real-time optimal control strategies are proposed concerning the distribution of the computation load among physical devices and in time-scale. A complex optimization task with high computational complexity (i.e., computation code and computation load) is decomposed into a number of simple tasks, and the corresponding agents are constructed for solving them. Adopting edge computing, a computing paradigm using the distributed computing resources of field IoT devices, these agents are implemented on the integrated local devices. This is an effective means for dealing with the network traffic caused by the centralized structure and rapid growth of IoT devices. The computation task of an optimization decision is further distributed into a number of steps, each performed at a sampling interval of the local devices. Adopting these two distribution schemes, the computation loads of all individual agents at each step were below 2000 FLOPs, allowing them to be handled by the typical smart sensors using simple optimization codes. A convergence acceleration method is proposed to speed up the convergence of the distributed optimization. Adopting this method, the number of iterations for each optimization decision was within 50, well below the convergence rate needed for optimal control with the typical time interval of minutes. The proposed agent-based optimal control strategy is also convenient and effective to deal with multiple components of different performances. The optimization considering such performance deviations could reduce the overall energy consumption significantly.
Information delays refer to the time delays in information exchange between devices integrated in communication networks. They could affect the performance of distributed optimal control, but are rarely concerned in HVAC field. This study investigates and quantifies the impacts of information delays on the performance of distributed optimal control strategies for HVAC systems through theoretical analysis and case studies, including a typical central cooling plant and a typical multi-zone air-conditioning system. The uncertain information delays are modelled by a Markov chain according to the characteristics of communication networks. Their impacts are quantified by comparing the performance of the distributed optimal control strategies involving the information delays with ideal performance. Results show that information delays significantly affected the convergence rate and control accuracy of the distributed optimal control strategies. These delays resulted in a difference in optimized cooling tower outlet water temperature of up to 0.6 K and a number of iterations of up to180 (about nine times than in ideal conditions). Test results indicate the necessity of considering the impacts of information delays when developing distributed optimal control strategies for HVAC systems.
To reduce the impacts of information delays on the performance of the distributed optimal control strategies, a delay-tolerant control method is proposed. It reduces the impacts of information delays through synchronizing the local optimization results used for convergence determination and adaptively resetting the step-size used for updating the Lagrange multiplier. The purpose of synchronizing local optimization results is to reduce the impacts of information delays on the accuracy of the optimization results. The purpose of setting the step-size adaptively is to reduce the impacts of information delays on the convergence rate. The computation load of the proposed method is 40 FLOPs (floating-point operations), which can be handled by typical smart sensors. Test results show that the proposed delay-tolerant control method could effectively reduce the impacts of information delays on optimization accuracy and convergence rate, thereby improving the energy performance of the distributed optimal control strategy under conditions where delays occur.
To further validate and evaluate the proposed distributed optimal control strategy, a hardware-in-the-loop simulator is constructed as a realistic test environment for the distributed optimal control strategies being implemented on real control devices. Being implemented on a wireless IoT sensor network integrated in the simulator, the applicability and the performance of the proposed strategy are validated and evaluated. The experience and test results show that the IoT sensing network has the capacity to implement the distributed optimal control strategy and handle the decomposed optimization tasks effectively. The energy performance of the proposed distributed optimal control strategy is almost the same as that using the perfect solutions.
To conclude, the proposed distributed optimal control strategies are applicable in the local IoT devices of field control networks and they are effective in improving system energy efficiency. For the current BASs, these strategies could broaden the applications of the optimal control of HVAC systems, by using the distributed computation resources of digital controllers integrated in field control networks. For the future BASs, they provide an approach to fully utilize the IoT-enabled field control networks for the optimal control of HVAC systems and support the development and applications of the emerging IoT technologies in the building automation industry.
|Rights:||All rights reserved|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: