Application of artificial neural network (ANN) and response surface methodology (RSM) for modeling and optimization of the contact angle of rice leaf surfaces

In this paper, a novel method for modeling and optimization of the contact angle of rice leaf surfaces by comparing the analytical result of the artificial neural network (ANN) and response surface methodology (RSM), to minimize the contact angle of the rice leaf surfaces. Three factors and three levels Box-Behnken design was used to analyze the external factors influencing the contact angle of the rice leaf surfaces, which consisted of temperature, humidity and the chemical pesticide concentration. By optimizing the designing parameters, the optimum input parameters were obtained. The best input conditions were temperature (25.96 °C), humidity (60.00%), chemical pesticide concentration (131.89 mg/L), and a minimum contact angle of 103.565° was obtained. Moreover, the ANN sensitivity analysis was used to study the weight of three external factors affecting the contact angle. Experiments showed the concentration of chemical pesticides to be the chief factor, as it has a weight is 55.61%, followed by humidity with a weight of 31.15% and temperature with 13.24%. Finally, the reliability of the RSM and ANN methods in modeling and predicting the contact angle was compared by analyzing the value of mean squared error (MSE) and determination coefficient (R2).The result showed that both models have better performance, and both RSM and ANN have better determination coefficients. The experiments also illustrated that the ANN method get better results than RSM in fitting experimental data, prediction and modeling.