Synergy of Self-optimizing Control and Real-time Optimization With Applications to Building Energy Systems
| dc.contributor.advisor | Li, Yaoyu | |
| dc.creator | Zhao, Zhongfan | |
| dc.date.accessioned | 2022-10-12T22:13:08Z | |
| dc.date.available | 2022-10-12T22:13:08Z | |
| dc.date.created | 2020-08 | |
| dc.date.issued | 2020-05-28 | |
| dc.date.submitted | August 2020 | |
| dc.date.updated | 2022-10-12T22:13:09Z | |
| dc.description.abstract | This dissertation research investigates on energy optimal operations of heating, ventilation and air conditioning (HVAC) systems for residential and commercial buildings, with investigations on two classes of approaches, i.e. Self-optimizing Control (SOC) based control structure optimization and Extremum Seeking Control (ESC) based real-time optimization, as well as their synergy. As the preliminary study, three existing ESC methods are compared, i.e. the conventional dither ESC, input-output correlation ESC (IOC-ESC), and proportion-integral ESC (PI-ESC). The primary goal for this study is to evaluate the performance robustness of each ESC to parametric variation. The three ESC strategies are first evaluated with simulation study using a Modelica model of an air source heat pump (ASHP) model. Then, laboratory tests are conducted with IOC-ESC and CON-ESC on a mini-split air conditioning system. Both simulation and experimental results show that PI-ESC archives fastest convergent speed, while IOC-ESC is most robust to parametric variation among three methods. For plant operation with multiple potential inputs for ESC design, an ESC integrated online input selection approach is proposed based on Hessian estimation and SVD analysis. The dimensionality of manipulated inputs can be reduced by pruning and/or combining the available manipulated inputs. Error bound analysis is conducted for the proposed input selection process. Then, the proposed procedure is validated with both numerical example and a Modelica based chilled-water plant model. The SOC based control structure optimization is studied from three perspectives, with the key focus of how to determine the optimal selection of controlled variables (CVs) with online or historical data. First, an online local SOC procedure is proposed by using the ditherdemodulation based gradient and Hessian estimates that have been developed in ESC, and the proposed method is validated with the Modelica ASHP model. Secondly, a data-driven global SOC method is proposed by applying the technique of sparse identification for nonlinear dynamics (SINDy) to historical data. For the static plant model thus derived, tracking the necessary condition of optimality yields the globally optimal solutions of CV selection in an analytical fashion. Such SOC framework is validate with the Modelica-based chilled-water plant operation. Then, this data-driven global SOC approach is extended to plant operation with wide range of disturbance variations which is subject to the changes of active constraint set. With the plant model obtained by the SINDy based procedure, a parametric programming enabled constrained optimization framework is utilized to derive the controlled variables for each region of different active constraint set. Again, simulation based validation is conducted with the Modelica based chilled-water plant model. | |
| dc.format.mimetype | application/pdf | |
| dc.identifier.uri | https://hdl.handle.net/10735.1/9501 | |
| dc.language.iso | en | |
| dc.subject | Self-organizing systems | |
| dc.subject | Heating ǂx Equipment and supplies ǂx Design and construction | |
| dc.subject | Ventilation ǂx Equipment and supplies ǂx Design and construction | |
| dc.subject | Air conditioning ǂx Equipment and supplies ǂx Design and construction | |
| dc.title | Synergy of Self-optimizing Control and Real-time Optimization With Applications to Building Energy Systems | |
| dc.type | Thesis | |
| dc.type.material | text | |
| thesis.degree.department | Electrical Engineering | |
| thesis.degree.grantor | The University of Texas at Dallas | |
| thesis.degree.level | Doctoral | |
| thesis.degree.name | PHD |
Files
Original bundle
1 - 1 of 1