Optimization of Counter-Rotating Cotton Stalk Puller Performance Using Response Surface Methodology and ANN-PSO Technique

Cotton stalks, a by-product remaining after cotton picking, have significant industrial value. However, their removal is difficult due to the deep taproot system, making stalk uprooting labour-intensive. This study focuses on optimizing the uprooting efficiency of a counter-rotating drum-type cotton stalk puller (CSP) using Response Surface Methodology (RSM) and an integrated Artificial Neural Network-Particle Swarm Optimization (ANN-PSO) approach. The developed tractor-operated machine consists of counter-rotating tapered drums with spring-loaded adjustable drum clearance mechanism to effectively grip and uproot stalks of varying sizes with minimal breakage, even under hard soil conditions. The machine’s operational and design parameters were considered as independent variables, whereas uprooting efficiency, plant breakage, and plants left in the field were considered response variables. An experimental CSP unit was evaluated under field conditions at three forward speeds (1.37, 1.67, and 1.95 km h^-1), four drum speeds (250, 300, 350, and 400 rpm), and three drum inclination angles (0°, 10°, and 20°). Optimization using RSM indicated that a drum speed of 332.5 rpm, a drum inclination angle of 8.36°, and a forward speed of 1.37 km h^-1 were the optimal operating conditions. Under these conditions, the optimum values of uprooting efficiency, plant breakage, and plants left in the field were 96.6%, 2.8%, and 1.1%, respectively, with individual desirability values of 0.97, 0.85, and 0.89, indicating good agreement between the predicted and observed responses. The ANN-PSO model predicted the optimal operating parameters as a forward speed of 1.37 km h^-1, a drum inclination angle of 7.89°, and a drum speed of 331.45 rpm. The observed and predicted values of uprooting efficiency under these conditions were 96.72% and 94.84%, respectively. The results demonstrated that both RSM and the integrated ANN-PSO approach effectively predicted and optimized the performance of the CSP, with high accuracy. The optimization study provides valuable insights into the optimal combination of design and operating parameters for achieving enhanced uprooting efficiency with minimal plant breakage.