Dynamic Programming Based Resource Optimization in Agricultural Big Data for Crop Yield Maximization

摘要

Precision agriculture uses sensor data, web data and stream data obtained from various sources such as soil moisture sensors, agro informatics websites and satellite images respectively, to design an agricultural big data framework for decision making and predictive analytics. As themajor resources of the agriculture, water, fertilizer and micronutrient resources are scarce and given the uncertain climatic conditions, an optimal and limited use of these resources become imperative. Added to these, the domain knowledge on the type of crop to be grown at a given location,soil type, weather conditions and market trend need to be analyzed in order to obtain maximum yield. The linear programming optimization model for resource management can minimize the resource requirement but have an exponential time complexity. It is also evident that the resources requiredvary from one stage to the other of the crop growth. Hence the proposed work uses Dynamic Programming based Resource Minimization Algorithm (DRMA) to optimize the water, fertilizer, micronutrients requirement based on the availability and requirement in each stage of the crop’s growth.The DRMA model optimizes water level, fertilizer and micronutrient requirements based on the type of soil, amount of rainfall and weather data. In addition, the proposed approach can also be extended by supplying the optimized values of the resources to an analytical model to decide on thetype of crop to be grown at any given location that would provide a better yield thereby giving higher Return on Investment (ROI).