Identification and Analysis of Key Parameters for Yarn Breakage in Direct Twisting Machines under Factory Environment Based on Random Forest Model

Abstract

The yarn breakage phenomenon during the twisting process of direct twisting machines in current factory environments is extremely common, resulting in raw material waste, frequent machine downtime, and increased energy consumption; therefore, a systematic analysis of the causes of yarn breakage is of great importance. Starting from the factory operating parameters of the direct twisting process, this study employs a random forest model, which excels at handling nonlinear relationships, to conduct an in-depth analysis of its operating patterns and influencing factors. The model was tuned via sample balancing, grid search, and random search, with the final configuration setting the maximum depth of decision trees to 10, the number of trees to 200, and evaluating the model using 5-fold cross-validation. The results show that the model achieves an F 1-score of 0.935 and a receiver operating characteristic–area under the curve (ROC-AUC) value of 1, demonstrating high accuracy. Based on feature importance analysis, the actual spindle operating speed and the package percentage were identified as the dominant factors contributing to yarn breakage, accounting for more than 80% of the total contribution, whereas the weight of traditional tension parameters was relatively low. This is attributed to the fact that tension parameters are strictly controlled within the process range during production, resulting in insufficient variation. The deviation between the actual spindle operating speed and the set speed significantly affects yarn breakage; when the relative deviation exceeds ± 0.08%, the risk of breakage rises significantly, as speed fluctuations induce instantaneous instability in yarn tension. Furthermore, the yarn breakage rate surges when the package percentage is below 30% or above 70%. Based on these analysis results, measures such as controlling the spindle speed within the Q ₁ – Q ₃ range of normal operating speeds, optimizing the anti-patterning angle parameters, and arranging regular manual inspections and guide device corrections are proposed to effectively reduce the yarn breakage rate.