Optimization of seeding performance of a mechatronic seed metering mechanism using integrated CANFIS-MOGA
integrated CANFIS-MOGA
Abstract
In row crops cultivation, uniform plant spacing is crucial for maximizing grain yield. The seed spacing uniformity of a mechatronic seed metering mechanism is influenced by its operational parameters i.e., forward speed, cell size and seed metering plate inclination. To achieve uniform seed spacing, optimization of these parameters is crucial. In this study, hybrid Coactive Neuro-Fuzzy Inference System (CANFIS) and Multi-Objective Genetic Algorithm (MOGA) was developed to optimize operational parameters. The developed mechatronic seed metering mechanism was tested in the laboratory using sticky belt test setup with selected variables. The forward speed, cell size and metering plate inclination were considered as independent parameters and miss index (Mi), multiple index (Mui) and precision index (Pi) were considered as response variables to develop the CANFIS model. The CANFIS predicted the response variables with R2 value of 0.99 during testing phase and 0.94 to 0.98 during training phase. Sensitive analysis of the independent parameters revealed that the forward speed is the most sensitive paramete, followed by cell size and seed metering plate inclination. The predicted results of the CANFIS were utilized as a fitness function in the genetic algorithm to obtain the optimal operational parameters of the mechatronic seed metering device corresponding to minimum Mi, Mui and Pi. The optimum parameters of seed metering device were found to be 3.24 km/h forward speed, 14 mm cell size and 50 degree seed metering plate inclination with Mi of 5.04 %, To verify the reliability of the simulated results of hybrid CANFIS-MOGA, the validation test was conducted and found relative error of 1.30 %, -3.12 % and 0.80 %, respectively between predicted and actual values of Mi, Mui, and Pi, respectively, thus demonstrating robustness and reliability of the developed hybrid model
