Predictions of greenhouse soil moisture using artificial neural network and wireless network sensing

Soil moisture is one of the main factors determining the better growth of plants which are widely well-received today, especially in greenhouses. Measuring the soil moisture and the environmental factors has high costs continuously and annually, in addition to being time-consuming. Therefore, one of the intelligent predictive tools that have a vast position in agricultural science is the neural network tool with the least amount of error. In this study, soil moisture and temperature percentage, light, ambient temperature, and humidity in a greenhouse located in northeastern Khuzestan were Measured and recorded during two seasons of winter and spring to control soil moisture by a moisture prediction map based on an artificial neural network. The results show an accurate forecast of soil moisture map in winter and spring between actual values that were measured and values that were predicted with the lowest standard error (1.12 and 1.71) and the highest coefficient of determination (R2) of 0.94 and 0.78, respectively, and the highest coefficient of determination were 0.87 and 0.93, respectively, by the artificial neural network in the experimental stage for winter and spring. Therefore, the remarkable accuracy in the prediction of soil moisture by this software shows its importance and high reliability in agriculture and greenhouses which makes it easier to control soil moisture and less moisture stress occurs for soil and the plant consequently.