ANFIS-Based Compensation of External Disturbances in the Cotton Seed Lintering Process

Abstract

The cotton seed lintering process is characterized by nonlinear behavior and significant sensitivity to external disturbances such as moisture content, impurity level, and raw material variability. These factors lead to instability in the output quality parameter, primarily seed fuzziness, resulting in increased production of non-conforming products. This study proposes an Adaptive Neuro-Fuzzy Inference System (ANFIS)-based approach to compensate for uncertainty and external disturbances in the lintering process. A multi-input ANFIS model is developed using four key input parameters: initial fuzziness, moisture content, impurity level, and drum rotational speed. The system generates a compensatory control signal to stabilize the process under varying conditions. Synthetic experimental datasets, reflecting real industrial behavior, are generated and used for training and validation. The proposed method demonstrates improved robustness and accuracy in maintaining desired output quality compared to conventional control approaches. The results indicate that ANFIS effectively mitigates the impact of disturbances, reducing variability in the final fuzziness parameter. The developed model can be integrated into intelligent control systems for cotton processing industries.