Intelligent Assessment of Wheat Yield through the Variable Potential of Seeds

Introduction. Evaluation of wheat seed quality is an integral part of the technological process of its production since it increases the yield. The yield is affected by many different factors. The evaluation methods are constantly being improved taking into account new factors, physical methods and technical means. Currently seeds and crops sowing quality intelligent evaluation methods are developing very rapidly. The electrophysical method allows evaluating the soil influence on seeds by the variable potential. Aim of the Study. The study is aimed at examining changes in the variable potential of wheat seeds of a known crop yields during seed swelling in solutions with different potassium and sodium ratios and creating a convolutional neural network to estimate potential crop yields through the variable potential and known potassium and sodium ratios. Materials and Methods. The studies were carried out using seeds of two varieties of spring wheat with different yields. To simulate soil quality, there were used solutions with different potassium chloride and sodium chloride ratios. The variable potential was measured using a device based on the data acquisition board LA50-USB. The yield was estimated using wavelet transform and deep convolutional neural network with ResNet groups. Results. There have been developed the experimental graphs of the variable potential change depending on the potassium and sodium ratio in a solution simulating soil quality. The neural network was used to classify the potential yield of wheat seeds through wavelet transforms of the variable potential, and potassium and sodium ratios. There has been compiled a table of neural network responses to test variable potentials. Discussion and Conclusion. The developed graphs of the variable potential change depending on potassium change in the external environment were compared with the results of studies by other authors. The results qualitatively coincide. The developed neural network can classify the potential yield of wheat seeds through the variable potential, and potassium and sodium ratios. The conducted study is useful for agricultural enterprises and farmers. The proposed methodology for assessing potential crop yields through variable potential and water extract will allow optimizing the process of potassium application to the soil.