Data-driven horticultural crop model for optimal fertigation management -a methodology description

Abstract

Fertigation has been proved a good agricultural practice for greenhouse crop production with respect to the efficiency of water and fertilizer use. but the frequency of irrigation and N fertilization, as well as their quantity, has important effects on crop’s high yield and quality. To further save the irrigation water and N fertilizer application meanwhile ensuring the maximum productivity, the rational fertigation management is indispensable. Crop growth model provide a useful tool to help determine optimal management practices. This paper illustrates the basic methodological approaches and structure for optimal fertigation management by using a data-driven model. The data-driven model is derived from the given data and nonlinear regression method is applied to set up the model. The model parameters are identified by ordinary least square method. After the model’s evaluation, a osculating value method is proposed to optimize the fertigation management combined with the data-driven model. The modeling and optimization algorithms, as well as the input and output interface, are implemented in standard c language, which is easy to be integrated into smart device like fertigation machine, to enhance the intelligence of device or controller. The experimental data from four crops including tomato, cucumber, sweet pepper, and strawberry were collected to assess the performance index of above methodology approaches, the preliminary results show that the system framework and the methodology adopted can be used to help farmer-level fertigation management as an alternative tool.