Jiri Cigler

Modeling and identification of a large multi-zone office building

Predictive control in buildings has undergone an intensive research in the past years. Model identification plays a central role in a predictive control approach. This paper presents a comprehensive study of modeling of a large multi-zone office building. Many of the common methods used for modeling of the buildings, such as a detailed modeling of the physical properties, RC modeling, etc., appeared to be unfeasible because of the complexity of the problem. Moreover, most of the research papers dealing with this topic presents identification (and control) of either a single-zone building, or a single building sub-system. On contrary, we proposed a novel approach combining a detailed modeling by a building-design software with a black-box subspace identification. The uniqueness of the presented approach is not only in the size of the problem, but also in the way of getting the model and interconnecting several computational and simulation tools.

Subspace identification and model predictive control for buildings

Model predictive controller presented in this article makes use of both weather forecast and thermal model of a building to control inside temperature. This, by sharp contrast to conventional control strategies such as weather-compensated heating control (heating-curve controller), enables utilization of thermal capacity of the building, thus minimization of energy consumption. The inside temperature can be maintained at desired levels independent of the outside weather conditions using modified formulation of predictive controller. Nevertheless, proper identification of the building model is crucial. The models of multiple-input multiple-output systems can be identified using subspace methods. The controller was tested on (and applied to) the real building and results were compared with a present heating control.

Strongly feasible stochastic model predictive control

In this article we develop a systematic approach to enforce strong feasibility of probabilistically constrained stochastic model predictive control problems for linear discrete-time systems under affine disturbance feedback policies. Two approaches are presented, both of which capitalize and extend the machinery of invariant sets to a stochastic environment. The first approach employs an invariant set as a terminal constraint, whereas the second one constrains the first predicted state. Consequently, the second approach turns out to be completely independent of the policy in question and moreover it produces the largest feasible set amongst all admissible policies. As a result, a trade-off between computational complexity and performance can be found without compromising feasibility properties. Our results are demonstrated by means of two numerical examples.

Subspace identification of poorly excited industrial systems

Most of the industrial applications are multiple-input multiple-output (MIMO) systems, that can be be identified using knowledge of the systems physics or from measured data employing statistical methods. Currently, there is the only class of statistical identification methods capable of handling the issue of vast MIMO systems - subspace identification methods. These methods, however, as all statistical methods, need data of certain quality, i.e. excitation of corresponding order, no data corruption, etc. Nevertheless, the combination of statistical methods and physical knowledge of the system could significantly improve system identification. This paper presents a new algorithm which provides remedy to insufficient data quality of certain kind through incorporating of prior information, e.g. known static gain or input-output feedthrough. The presented algorithm naturally extends classical subspace identification algorithms, that is, it adds extra equations into the computation of system matrices. The performance of the algorithm is shown on a case study and compared to current methods, where the model is used for an MPC control of a large building heating system.

Predictive control oriented subspace identification based on building energy simulation tools

Even though modern control has emerged in numerous control applications, a building automation is still a field where the position of the classical control is almost exclusive. The main reason is that for the synthesis of a predictive controller a decent model for control is needed. In the field of building climate control, it is still problem to obtain a model of large building in an explicit form suitable for control. Most of the approaches either use building modeling software to get detailed model, which is unfortunately in implicit form; or the model is built-up as a first principle model, which usually ends-up as an extreme simplification of the reality. In this paper, a building model identification procedure is presented, wherein the building model is built-up as a first-principle model using a simulation software (detailed, precise, however in implicit form), and then a state-space model is identified by means of subspace identification methods. The main focus of the paper lays on a case study of a large office building, and the entire process of its identification.

Optimization of predicted mean vote thermal comfort index within Model Predictive Control framework

Recently, Model Predictive Control (MPC) for buildings has undergone an intensive research. Usually, according to the international standards, a static range for the air temperature represents the thermal comfort which is being kept making use of MPC while minimizing the energy consumption. On contrary, this paper deals with the optimization of the trade-off between energy consumption and Predicted Mean Vote (PMV) index which, opposed to the static temperature range, describes user comfort directly. PMV index is a nonlinear function of various quantities, which makes the problem more difficult to solve. The paper will show the main differences in MPC problem formulation, propose a tractable approximation strategy and compare the control performance both to the conventional and typical predictive control strategies. The approximation of PMV computation will be shown to be sufficiently precise and moreover, such a formulation keeps the MPC optimization problem convex. Finally, it will be shown that the proposed PMV based optimal control problem formulation shifts the savings potential of typical MPC by additional 10% while keeping the comfort at a desired level.

On predicted mean vote optimization in building climate control

Low energy buildings have been attracting much attention lately. Most of the research is focused on the building construction or alternative energy sources. Recently, there has been an intense research in the area of Model Predictive Control (MPC) for buildings. The main principle of such a controller is a trade-off between energy savings and user welfare making use of predictions of disturbances acting on the system (ambient temperature, solar radiation, occupancy, etc.). Usually, the thermal comfort is represented by a static range for the operative temperature according to the international standards. By contrast, this paper is devoted to the optimization of the Predicted Mean Vote (PMV) index which, opposed to the static temperature range, describes user comfort directly. PMV index, however, is a nonlinear function of various quantities, which makes the problem more difficult to solve. The paper will show the main differences in MPC problem formulation, compare the control performance both to the conventional and predictive control strategies, point out that the proposed optimal control problem formulation shifts the savings potential of classical MPC by additional 11% and finally, the quality of the fulfillment of the thermal comfort will be addressed.

Usage of spot market prices prediction for demand side management

This paper discusses the issue of demand side management, in particular control of the output power of heat pumps based on the spot market electricity price. The main presumptions are the ability of controlled HVAC system to shift energy load on the customers side and sufficient credibility of an energy price prognosis on the electricity providers side. The paper first presents current situation in the Czech republic with electricity tariffs legislations, which have to be followed so that the proposed method is applicable in practice. It is shown that for successful implementation, it is required to have an energy load model, model of the accumulation as well as parameters of the heat pump. For the energy load model, statistical black box methods are used, while the other models are based on first principles. The models are together with the prediction of spot market price used within model predictive control framework resulting in cost savings higher than 10%.

Application of knowledge-based control on an antibiotics production fermentation process

Modernisation of control systems of industrial biotechnology processes to the current technology levels presents a challenging task because nowadays, there is a bigger stress put on the automatisation of the production process so that the possible human fault is minimised. Therefore in this paper, the process of sequential modernisation of the control system from “almost fully controlled by process operators” to “almost supervised control” is presented. A decision support system based on various knowledge about the process is developed to support actions performed by process operators. The system brings improvements in antibiotic production operations and reduces manufacturing variance.