Abstract
Real time parameter estimation of dynamic models operating in closed loop control is a crucial issue to implement industrial adaptive controllers. Parameter estimation must be seen as one of the key elements to solve a system identification problem, which involves also an experiment design, the selection of a model structure, and the model validation. This paper describes some approaches to compute the transfer function parameters of time-varying systems under closed loop control. To point the limitations and advantages of each method, with focus on robustness and quality of the model estimates, the techniques are applied to compute the parameters of a time varying discrete system, with known structure, under PI- Proportional-Integral control.

Abstract
Improvement of crop quality and yields is a demand in modern greenhouse production systems. Also, production costs must be kept as low as possible to guarantee market competitiveness. The achievement of these goals implies the use of complex management and control systems to regulate, in an efficient way, a large amount of interactive physical variables. Recent developments in hardware and software tools, namely microprocessors and microcontrollers, lead to the integration of complex control and management tasks in agricultural exploitations. In this paper is presented a data acquisition and control network that was implemented with the aim of being applied to any agricultural environment. The network has three main operating levels. At the lower level, a set of remote microcontroller stations perform data acquisition and radio frequency transmission to a collecting and control station. The control station, which generates actuating signals, is linked to a higher-level network based on PC's. The management and supervision of the entire greenhouse system is performed at this lever. Also, the results achieved with its application to the environmental control of a set of greenhouses located in the north of Portugal are described. The proposed architecture is now being installed in several commercial exploitations in order to evaluate its performance and introduce any improvements required by the growers. After this phase, the network will be available commercially through a joint Venture between UTAD University and a Portuguese greenhouse constructor.

Abstract
A multivariable predictive controller was implemented to regulate the air temperature, humidity and CO(2) concentration for a greenhouse located in the north of Portugal. The controller outputs are computed in order to optimise the future behaviour of the greenhouse environment, concerning the set-point accuracy and the minimisation of energy inputs. This is accomplished by using an optimisation module that minimises a cost function proportional to the sum of the squared errors between the simulated and desired outputs plus the square of the incremental and absolute energy inputs over a prediction horizon of one hour. Since the controller must be able to predict the greenhouse environmental conditions over the specified time interval, it is necessary to use mathematical models that describe the greenhouse climate, as well as to predict the outside weather. The experiments showed that second order ARX and tenth order ARMA models are well suited to simulate the inside and outside climate conditions, respectively. Since the model parameters are time-varying, recursive identification techniques were applied to estimate in real-time their values. The models employ data from the air temperature and humidity, inside and outside the greenhouse, solar radiation, wind speed and control inputs. To minimise the cost function a sequential quadratic programming method was used to solve the constrained optimisation problem. The results achieved with the proposed controller proved to be suitable for this application. Moreover, the controller performance, when compared to other control techniques such as commercially available PID controllers, was greatly improved.

Abstract
Greenhouse control computers are an essential part of modern greenhouse operation. Climate, irrigation and nutrient supply must be controlled, in an economically way, to produce the best crop conditions. Current research on CO2 enrichment and optimal growth strategies implies the use of powerful tools, either based on hardware or software. This paper describes the implementation of a distributed data acquisition and control system for computerised agricultural management systems. To accomplish with the emergent and future tendencies in this area the network uses different communications platforms to achieve low-cost, flexibility, and functionality. The techniques and tools, that provide to the user a transparent, friendly and intuitive Graphical User Interface will be also presented. Copyright (C) 1998 IFAC.

Abstract
A real-time parameter estimator for the climate discrete-time dynamic models of a greenhouse located at the North of Portugal are presented. The experiments showed that the second-order models identified for the air temperature and humidity achieve a close agreement between simulated and experimental data. The real-time data acquisition and the recursive identification techniques implemented are used in the simulation and design of an adaptive climate controller to achieve set-point accuracy and minimisation of the energy inputs.

Abstract
A fuzzy identification model and fuzzy logic controller was developed aiming the environmental control of an agriculture greenhouse. The fuzzy identification of the process was performed by the analysis of the data collected either during normal operation, as well in reaction to random generated actuating signals on the heating, ventilation and CO2 injection systems. A comparative study has been realized between fuzzy and physical models. Using the fuzzy model it was possible to implement an accurate Generalized Predictive Control (GPC) strategy in order to regulate the environment and to minimize energy consumption. The optimal setpoints were computed by means of balancing the benefits associated with the marketable produce against the costs associated with its production. The calculations use growth, photosynthesis and climate models. This work describes the practical development of an fuzzy controller that memorize the optimal strategy, executed by the GPC, to regulate the temperature and the CO2 concentration of the greenhouse inside air.