<![CDATA[This study aims to predict the quality of the soldering process using a wave soldering machine by utilizing the Sugeno model Fuzzy Inference System (FIS) and Adaptive Neuro-Fuzzy Inference System (ANFIS) methods. Soldering is a critical stage in PCB (Printed Circuit Board) production, where its quality is influenced by parameters such as solder temperature, conveyor speed, and flux volume. Traditional approaches such as visual inspection are considered less effective because they are prone to human error. Therefore, this study proposes the use of Sugeno FIS and ANFIS to model the non-linear relationship between process parameters and soldering quality, which is measured through Defect Per Opportunity (DPO). Data were obtained from the actual production process and processed using MATLAB. Sugeno FIS was applied with fuzzification, rule making, and defuzzification, while ANFIS combines neural networks with fuzzy logic for data-driven optimization. The results showed that both models were able to predict DPO with high accuracy, indicated by very small Root Mean Squared Error (RMSE) values (0.00179 for FIS Sugeno and 1.31597 × 10⁻⁶ for ANFIS). ANFIS excels in capturing non-linear complexity, especially in conveyor speed variations. Simulations using SIMULINK prove the effectiveness of this model in real-time prediction. These findings provide an innovative solution for the electronics industry to improve soldering quality automatically. Keywords - Wave soldering, FIS Sugeno, ANFIS, quality prediction, DPO.]]>
Copyrights © 2025
