Micro-climate optimal control for an Experimental Greenhouse Automation Lala H. Rajaoarisoa 1 , Nacer K. M’Sirdi 2 and , Jean-Franc ¸ois Balmat 3 (1) EMDouai, IA, F-59500 Douai, France (2) LSIS, Avenue Escadrille Normandie-Niemen, 13397 Marseille Cedex 20 (3) LSIS, Universit´ e du Sud Toulon Var, Facult´ e des Sciences et Techniques, 83957 LA GARDE Cedex lala.rajaoarisoa@mines-douai.fr, nacer.msirdi@lsis.org, balmat@univ-tln.fr Abstract The goal of a greenhouse process culture is to give plants a suitable environment so they can develop prop- erly. On the other hand, this kind of culture allows minimizing emissions and production costs. To reach these objectives, the need an optimal control arises. However, it is well known that agricultural greenhouse is a very complex system, composed of elements that can interact and exchange of energy between them and with their environment. So, the setting and tuning of greenhouse climate controllers is by no means an easy or standard procedure. In this paper, we develop one practical approach to design an optimal controller for this kind of system. The design strategies are based on stated stability and stability theorems for hybrid systems. The questions related to control techniques are discussed to show in this way the complexity of the controller design of a class of the studied system. Index Terms—Greenhouse plant, optimal control, stabil- ity, controllability, controller design. I. . Introduction F OR several years now, the modeling of greenhouses became the subject of many research works [18], [5], [8], [10], [7], [16], [11], [17], [1]. These works consist mainly in the study of phenomena energy related to changes in the internal climate of greenhouses. Indeed, the goal is generally to optimize the quality and produc- tion efficiency by implementing a suitable control. Let’s recall that greenhouse plant is an agricultural and energy complex biological system. The production process of an agricultural emission can be represented by two main parts. The first one is composed of two separate steps. The first step is used to check the nature of the production and the type of culture and the second step is the system calibration to the optimal conditions desired to develop plants to grow. The second part shows the various actions required for the analysis of a greenhouse. The first action is necessary for internal states control of the greenhouse itself, including temperature, humidity and CO 2 levels. The second action is to manage the needs of plants. And finally, the third activity concerns the crop and its distribution. According to the intrinsic greenhouse’s features, the setting and tuning of greenhouse climate controllers is by no means an easy or standard procedure. The large number of greenhouse controller settings makes it difficult to foresee the influence on the results and the costs involved. One uses optimal control to reach the resulting complex production system. Of course greenhouse climate management can be significantly improved by implementing advanced con- trollers designed by using optimal control theory [9], [6], [19], [2], so the controller design must be implementable in practical case and requires the model to be smooth. But according to this outcome, in this paper, we develop one practical approach to design an optimal controller for this kind of system. Moreover, the knowledge of control design methods is limited in process industry and it is highly desirable to have simple design methods to deal with control’s problem. So, in the following, we propose a new straightforward design methods based on stated stability and quadratic stability results for hybrid systems. If certain conditions are satisfied we can ensure that the closed-loop system becomes stable or quadratically stable. This paper is also organized as follows. Firstly, we describe the hybrid dynamic model of our experimental greenhouse. In the second part, we present the methodology to deal with control techniques problem of a hybrid complex system. In the third part, we discuss on the stability and con- trollability properties of each sub-model. Finally, in the last part, we conclude and give some perspectives for a further work.