Sign In | Sign Up

Actions

WP3 - Optimization & Control

Objectives

Develop and test via simulations a fully-functional real-time EPB optimization & control tool.

Tasks

BO&C Algorithms Development

For all three demonstration buildings the following will be performed:

  • Based on the integrated EPB models of tasks T2.2 and T2.3, an offline globally-PEBBLE MPC methodology will be developed. Dynamic programming tools will be used towards such a purpose: based on the integrated predictive EPB models developed in T2.2.-T2.3 (i.e. integrated EPB and user-behavior models as well as the weather forecast model), the dynamic programming approach will be used to offline compute the optimal long-term (>24h) BO&C actions by searching in the space all available BO&C actions and evaluating the performance of candidate sets of BO&C actions using the integrated predictive models of T2.2-T2.3. Different EPB/user behavior/weather scenarios will be simulated (including a large variety of weather and occupancy/demand variations as well as a large variety for EPB’s physical uncertain parameters) and the corresponding optimal BO&C actions will be computed using the above-mentioned dynamic programming/MPC methodology. In this way, a large set of data containing a wide range of different EPB/user behavior/weather scenarios and the corresponding BO&C optimal actions will be created.
  • Based on (a) the set of data – created at the previous step – containing the different scenarios and the respective BO&C optimal actions and (b) the BO&C system requirements as identified in task T3.1, an AMPC system for BO&C will be constructed, so that (a) the overall mismatch between the AMPC control actions and the respective BO&C optimal control actions for each of the aforementioned scenarios is minimized, and (b) the AMPC meets the requirements of T3.1. A standard approximate dynamic programming approach will be adopted [Abu-Khalaf2005, Balakrishnan1996, Bertsekas1996, Lendaris2002, Murray2002] for the construction of the AMPC: such an approach employs experimentation with different combinations of neural and fuzzy approximators in order to efficiently approximate the optimal (MPC) system performance by an AMPC one that meets the computational and other constraints defined in T3.1.
  • Finally, a CAO system will be constructed for the on-line adaptive optimization of the AMPC tunable parameters. Although, the CAO systems developed in the past are quite generic and can be readily deployed to general control systems, special attention will be given within this task for the proper functioning of the user-interfacing system that is used to provide CAO with operator-imposed constraints and requirements. Such an interface will allow the operator to impose hard or soft constraints on the actions of the CAO automated optimisation system. More precisely, this interface will allow the operator to define:
    • Maximum and minimum allowable values that some or all of the tunable parameters or available measurements (e.g., temperature or humidity within specific rooms, energy consumed for certain periods of time) while the automated optimization process is active.
    • Conditional constraints, objectives and requirements, e.g., statements of the from “if the temperature at room A is less than x and the temperature at room B is less than y, then the HVAC air flow rate should should not exceed z; the definition and handling of such conditional constraints, objectives and requirements will be made possible by using available methodological tools in the design of human-machine interfaces [Ezzedine 2001].
Task 3.2 will be concluded by delivering the full-functional BO&C systems (delivered in the form of a software prototype along with all appropriate I/O modules) for the thee demonstration buildings.

BO&C System Assessment and Initial Design using Simulation Experiments

Simulation is a valuable tool for the cost-efficient, safe, time saving testing of BO&C systems. Moreover, someextreme situations that occur rarely in practice (extreme demands and weather conditions, device failure etc.) can be considered at command. Therefore, simulation is an indispensable tool when designing relatively complex systems with potential risks. Within this task, performance assessment of the integrated EPB+PEBBLE BO&C system for the three PEBBLE Demonstration Buildings will be performed for suitable scenarios of demand/weather conditions (which will be different than the scenarios used for the AMPC construction in task T3.2, but anyhow will be based on actual measurement data from the demonstration sites). Such a performance assessment will allow for:
  • Debugging and testing of the proper functioning of the BO&C system.
  • Assessment of the overall PEBBLE BO&C system performance for a wide range of demand and weather conditions (including both regular as well as atypical situations).
  • Check whether there is any need for add-ons on PEBBLE system.
  • Finally, provide an initially set of optimized set of AMPC tunable parameters by applying the CAO optimization system on-line (using the simulated building operations).
During the above-mentioned simulation assessment, different scenarios of user/occupant behavior and a variety of user-imposed constraints and requirements will be investigated. Task T3.3 will be concluded by issuing Deliverable 3.3: Simulation Assessment of PEBBLE System for the Three Demonstration Buildings.
© 2010-12 PEBBLE Consortium Partners. All rights reserved. Web-site developed by Katsigarakis Kyriakos  W3C Valid W3C Valid
Summary Impact Technical Approach Work Packages Project Info
News Events
TUC FIBP RWTH
Scientific Papers Conference Publications Dissemination Materials Deliverables Others
TUC FIBP RWTH TUGRAZ CSEM ARMINES SBC