تکنیک های پیش بینی شده کنترل برای انرژی و مدیریت کیفیت محیط درونی در ساختمان

The aim of the present paper is to present a model-based predictive controller, combined with a Building Energy Management System (BEMS). The overall system predicts the indoor environmental conditions of a specific building and selects the most appropriate actions so as to reach the set points and contribute to the indoor environmental quality by minimizing energy costs. The controller is tested using a BEMS installation in Hania, Crete, Greece.

During the last decades the contribution of Building Energy Management Systems (BEMS) to energy efficiency, improvement of the indoor comfort and environmental quality during a building's operational phase is well recognized. Advanced control techniques based on artificial intelligence (neural networks, fuzzy logic, genetic algorithms, etc.) and distributed control networks offer numerous benefits towards that direction [1], [2], [3], [4] and [5]. Building energy management and online control systems are reactive to the climatic conditions, building operation and occupancy interventions. Predictive control in conjunction with BEMS on the other hand uses a model to estimate and predict the optimum control strategy to be implemented [6]. While the online control systems can react only to the actual building conditions [7], a model-based predictive control can move forward in time to predict the buildings' reaction to alternative control schemes. Therefore different control scenarios can be evaluated based on suitable objective functions, and create a control state space that corresponds to a building's performance space [8]. A model can be either a “black box” or a “physical” model. In the “black box” or non-physical model approaches, self-learning algorithms, reinforced learning [9] or neural networks [10] are some of the methodologies found in the literature. The benefits of the mentioned approaches are low computational time and the fact that they do not require any specific building modeling expertise, while their limitations are (i) the fact that neural networks require reliable training data that may not be available and (ii) self-learning algorithms cannot move beyond the limits of their experience. When physical models are utilized, the expert has the opportunity to understand the cause-and-effect relationship between the various building components, the control strategies and the climatic conditions. The physical models approach can use stochastic mathematical models [11] or simulation-assisted predictive control [12]. Some physical models though require high computational skills and effort. In the present work a bilinear model-based predictive control is utilized in conjunction with BEMS, so as to achieve optimum indoor environmental conditions while minimizing energy costs. The bilinear modeling procedure is selected as it is the simplest extension of linear modeling and offers simplicity in the prediction algorithms' calculation procedure. The paper is organized in six sections. Section 2 includes a short description of a building and the installed BEMS. Section 3 incorporates the bilinear model analysis and the identification procedure. Section 4 analyses the predictive control strategy, while Section 5 presents shortly the graphical environment of the predictive control scheme. The experimental analysis including comparison between real and simulated measurements and discussion is presented in Section 6. Finally, Section 7 accumulates the conclusions and discusses issues for future research and development.

The present BEMS supported by a model-based predictive controller follows the initial design specifications and objectives. The system's response to the environmental variables' fluctuations is fast and stable. Considerable variations between the predicted and the real values are observed for the carbon dioxide concentration. This can be attributed to the fact that the window opening is small in comparison with to the building's volume and cannot contribute significantly to regulate the people's contribution and disturbances satisfactorily. The CO2 concentration though is kept close to the requested set point. Finally the controller's performance is quite satisfactory and selects the optimum solutions based on the energy consumption and the set-point proximity by satisfying the performance index J. It remains to future investigations to expand the performance index J to include indoor comfort requirements and improve the predictive control algorithm using different prediction techniques.