摘要：The importance of the factors that influence urea hydrolysis rate was analyzed to reveal the mechanism of urea hydrolysis and improve urea utilization efficiency. BP-K and DE-K prediction models were established based on the traditional back-propagation (BP) neural network model and the BP model optimized by the differential evolution algorithm (DE). The two models designed to predict the urea hydrolysis rate were validated with the measured value. The input variables in the two models were soil temperature, nitrogen amount, and moisture content, and the output variable was soil urea hydrolysis rate. Results showed that the BP-K and DE-K models possessed high accuracy and could be used to predict the urea hydrolysis rate. The simulation effect of the DE-K model was better than that of the BP-K model. The importance of the factors that affect urea hydrolysis was studied based on information on weight and the threshold value of DE-K model. The relative importance of temperature, nitrogen amount, and moisture content on urea hydrolysis was 65.87%, 25.69%, and 8.44%, respectively.
