Artificial intelligence techniques and near-infrared spectroscopy for nitrogen content identification in sugar cane crops

The strong rising on demand for agricultural crop quantity and quality, beyond the growing concern of non-point pollution, are requiring higher levels on agricultural production systems efficiency and environmental safety. Measurement of the nitrogen (N) on cultivation is one successful method of increase the agricultural production systems. Nitrogen application can increase productivity, decrease environmental impacts and also reduce costs by using the correct amount of nitrogen fertilizers. Recently, optical sensors applied to nitrogen measurement are attracting interest of several researches as a technique to enhance the productivity of sugar cane plants. However, the accuracy of measurements of reflectance still needs to be improved. This work proposes a new identifying approach based on near infrared reflectance (NIR) spectroscopy real time sensor using artificial intelligence techniques in order to improve the accuracy of the nitrogen measurements in sugar cane. An optical sensor is used to estimate the amount of nitrogen by reflectance measurement of sugar cane plants in the early stages of growing and the data is post-processed using intelligent systems. The results obtained using self-organizing map (SOM) presented better perform than other techniques for nitrogen content identification in sugar cane crops confirming the efficiency of intelligent approach in the real time.