科研之友
微信
新浪微博
Facebook
分享链接
摘要
In greenhouse vegetable production it is important to optimize and control the environmental management using dynamic models. The development and use of mechanistic models is expensive and time consuming. Black box models based on measurements of inputs and outputs are a promising approach for studying complex and nonlinear systems. In this work we have studied and generated neuro-fuzzy models to predict the behavior of temperature and relative humidity in two greenhouses. Input variables were: temperature, relative humidity, global solar radiation and wind speed and direction, measured outside the greenhouse. The output variables were temperature and humidity measured inside the greenhouse. The sampling time was every minute. Several neuro-fuzzy models for temperature and humidity were evaluated, using the neuro-fuzzy model ANFIS (Adaptive Neuro-Fuzzy training of Sugeno-type Inference System), available in the Fuzzy Logic Toolbox of Matlab. The methods of grid partition and subtractive clustering were used to generate the fuzzy inference system. Several empirical partitions of data were analyzed, as well as three types of membership functions (Gaussian, generalized Bell and Trapezoidal) and the constant and linear output membership functions. Also several training times were tested. Two sets of data were used, which were collected in two greenhouses with natural ventilation, located at the Universidad Autonoma de Chapingo and Universidad Autonoma de Queretaro. The analysis of the results showed that the neuro-fuzzy models acceptably predict the weather behavior inside the greenhouse.
- 出版日期2010-11
