Intelligent Automation System for Critical Analysis of Sewing Thread Breakage and Effects to Enhance Productivity
Alam, S M Masum
MetadataShow full item record
Sewing thread Breakage is one the leading cause for damage the garments productivity all over the world. In Bangladesh, various factories face the similar problem. The factors that are affecting the breakage are depended on the properties and quality of the yarn. Except for the thread quality, some other features also involve to breakage the thread. Types of the fabric, the weight of the unit area, the thickness of the fabric, thread count, strength, elongation % of the sewing thread and Needle number, machine speeds, yarn tension, stitch per inch are major factors for thread breakage. An appropriate prediction mechanism technique can apply to significantly reduce this problem. In this work, we have proposed an intelligent system that can make an effective prediction of a possible sewing thread breakage. Eleven (11) features and nine (09) well known supervised machine learning algorithms we have applied to analyze the prediction accuracy percentages. The overall process has been categorized into four stages. Stage 1: we have provided a comprehensive literature review where we summarize various related machine learning algorithms. Stage 2: Thread breakage data have been collected by the survey questionnaires and practical observations from the different apparel industries of Bangladesh. Stage 3: The feature factors data have been filtered and processing. Finally, fed the data to suitable machine learning algorithms to determine the predictive model and accuracy percentage of sewing thread breakage. ANN (artificial neural network) multilayer perceptron scaling accuracy specified for 11 functions, Comparative supervised machine learning, decision tree (J48), random forest, stochastic gradient, and sequential optimization have 100% accuracy, and with other classification algorithms such as Perceptron Multilayer, accuracy (99.80%) and logistic regression (98, 40%). Unlike Naive Bayesian Naive Bayesian (94%) has been updated. In Bangladesh, most companies are unaware of the cost of refining yarn and remanufacturing productivity. The main reasons for thread cutting are the lack of skills associated with production and inexperienced labor This intelligent system can predict the condition of a seamstress by measuring simple information that plant personnel find useful. Finally, it will improve productivity and the economy as a whole.
- M.Sc Thesis/Project