Artificial Neural Network as a Prediction Tool in Agricultural Variables

Abstract

Artificial Neural Network (ANN) technology is a form of artificial intelligence that learns byprocessing representative data patterns through its internal architecture. ANN technology oftenoffers a superior alternative to traditional physical-based models, and excel at uncovering patterns orrelationships in data. It is also a powerful non-linear estimator which is recommended when thefunctional form between input-output is unknown or it is not well understood but it is believed couldbe nonlinear.<br><br>This paper show two applications of ANN for forecasting solar radiation and available dam storage.In the first case ANN provided a good accurate forecasts for all period with an average square errorof 0.05% in the prediction. For the second case, ANN provide relatively accurate estimates of wateravailability one month into the future in the Purisima dam located in the state of Guanajuato. The results suggest that additional input variables such as runoff, cropping patterns and groundwaterextractions may be necessary to increase ANN forecasting accuracy.<br><br>This feasibility study demonstrates that ANN technology has the potential to serve as a highlyaccurate forecasting tool. Moreover, ANN technology can continuously be updated, as new databecome available, increasing its forecasting ability.