Abstract
(Englisch)
|
The project EDIFICIO aimed at optimising the control system for building services (heating, cooling, ventilation, blinds, artificial lighting), through the integration of several advanced information techniques into a global control system. The system elaborated by the project is characterized by the following specificities:
- the use of fuzzy logic rule bases for describing the operation of partial control systems (heating, cooling, etc);
- the use of self-adaptive techniques, allowing a continuous adaptation of the system to changeable conditions (weather, building characteristics, user's behaviour), and using essentially artificial neural networks;
- the use of a global 'cost function' allowing the long term adaptation of the integrated control system, through the minimization of that function by genetic algorithms.
The techniques used in the project are inspired by several other projects carried on by the EDIFICIO partners. In Switzerland, the projects 'Predictive Stochastic Control' (LESO-PB/EPFL), 'DELTA, A Blind Controller Using Fuzzy Logic' (LESO-PB/EPFL, Zumtobel Licht, Landis & Stäfa, Technical University of Vienna) and 'NEUROBAT, A Self-Adaptive Heating Controller' (LESO-PB/EPFL, CSEM, Sauter) have provided an essential background knowledge. Moreover, at LESO-PB/EPFL, several developments have been done in common with another UE-Joule project, the research project Smart Window (the coordinator of that latter project, TNO, was represented by the same person in both projects).
The EDIFICIO project included a first phase for the development of control algorithms by the partners, and a second phase for experimental tests (by simulation and by measurement on three full-size occupied building: one in South of Italy, one in Switzerland, and the third one in Finland, in order to cover a complete spread of European climates). The LESO-PB/EPFL was mainly involved in the development of algorithms for the control of blinds and artificial lighting and for the global adaptation, and in the experimental tests on the LESO building.
Work programme of the EDIFICIO project
- Control assessment: standards for the design of smart control strategies, performance criteria, operative conditions, and best areas of application;
- Algorithm development: individual predictive/adaptive neuro-fuzzy control algorithms for heating, cooling, lighting, indoor air quality, ventilation, and integration into a multi-functional control system;
- Experimental tests: full-size experiments in occupied office buildings located in North (Helsinki), Centre (Lausanne) and South (Catania) Europe, in order to assess energy savings and indoor comfort under different weather conditions, building occupancy and user behaviour;
- Dissemination: contacts towards manufacturers and practitioners, workshops and conferences, INTERNET, and elaboration of a pilot BEMS network design for a real building.
Software architecture for the control algorithms
The control algorithms are implemented as three nested levels:
- Level 1 control loop to manage the physical devices, which are different for each testing building; the control modules translate command variables (e.g. power [W], ACR [m3/h], blind position, slats orientation) into electrical signals for actuators.
- Level 2 control loop to realise a global strategy to simultaneously optimise all the devices (heating, cooling, ventilation, lighting, shading), based on Fuzzy controllers which receives as input information coming from sensors, predictive modules and user wishes.
- Level 3 to optimise the membership functions of the Level 2 Fuzzy controllers, through a continuous adaptation of the control parameters based on an overall cost function, taking into account all the constrains imposed by the technical equipment (energy consumption), changing of operative conditions, discomfort and reaction from the user.
Project results
Experimental tests at LESO, VTT, TNO and Conphoebus test buildings have been carried-out in 1999-2000. Results obtained on the LESO building have been used to improve the control algorithms and have shown that the EDIFICIO controller leads to 40% of energy savings in winter, 7% in mid-season and 18% in summer compared to a simple proportional controller. Overall, It has saved 24% of energy consumption in comparison with the conventional controller over the whole experiments period (27 January 2000 - 25 July 2000).
Further simulation runs have been carried-out by partners to check the performance of the complete system in other operating conditions, and with variants of the control algorithms. This has demonstrated clearly the adavantages of the EDIFICIO system in a variety of climate, building types and installed devices (with or without cooling, ventilation, shading control etc).
|
