To face a market increasingly competing and conditioned by increasingly severe standards of quality, the cultures under greenhouse experience a significant development. The systems of greenhouse production then become sophisticated considerably and inordinately expensive. This is why, the greenhouse designers who want to remain competitive, must optimize their investment by a great control of the production conditions [3]. New techniques of climatic control emerged, by which the use of artificial intelligence, such as fuzzy logic, neural networks, Neuro-Fuzzy are methods of topicality which are used for the development of mathematical models intended for the field of climatic management. From this report we have undertaken this study whose principal objective is the Neuro-Fuzzy modelling of a nonlinear system (the greenhouse). The latter characterizes the operation of the complex system which the greenhouse constitutes. The identification which is in the center of this step is a process of search for a mathematical representation which minimizes the variations of the real system compared to the modelled system. The tests are made on a sequence of May 1991, experimental data collected on the site of the INRA, Bioclimatology Avignon (France). The model obtained accurately represents the greenhouse and ready for the use within the framework of an on-line control [2].
