The objective of this paper is to investigate the predictability of the warp breakage rate from a sizing yarn quality index using a feed-forward back-propagation network in an artificial neural network system. In order to achieve the objective, a series of trials is conducted. An eight-quality index (size add-on, abrasion resistance, abrasion resistance irregularity, hairiness beyond 3 mm, breaking strength, breaking strength irregularity, breaking elongation, and breaking elongation irregularity) and warp breakage rates are rated in controlled conditions. A good correlation between predicted and actual warp breakage rates indicates that warp breakage rates can be predicted by neural networks. A model with a single sigmoid hidden layer with four neurons is able to produce better predictions than the other models of this particular data set in the study.There are numerous factors that can affect the performance of warp yarns in weaving. These factors can be broadly classified into the following three classes: yarn quality, condition of warp preparation, and loom actions and conditions [1,5].The weaving performance of a yarn is generally expressed in terms of warp breakage rate in weaving. Numerous attempts have been made in the past to predict warp breakage rates either from data of conventional tests of sized yarns or from test data obtained on specially designed instruments [2-4, 6 -7, 9, 13-15]. The approaches used in the past to estimate warp breakage rates can be broadly classified into empirical, statistical, and instrumental [2]. The phenomenon is so complex that it is difficult to form a clear picture of the breakage mechanism involved. The problem that remains is a correlation of the laboratory evaluation to the actual performance of a warp on a loom.An artificial neural network is composed of simple elements operating in parallel, which are inspired by biological nervous systems [8]. As in nature, the network function is determined largely by connections between the elements.Artificial neural networks are characterized by the fact that they make it possible to work without an analytical (explicitly formulated mathematical) model, whose basis is data. Recorded data are used to train a model with a learning capability in such a way that it reproduces reality (represented by the data) as faithfully as possible. Both the quantity and quality of the data are important. The objective is to model the input/output behavior of the system in question.A neural network is trained so that a particular input leads to a specific output. A large number of neural models now exist, and each of these models is available in various forms. A first criterion as to which method is most suitable results from the goal to be achieved.In this study, we investigate warp breakage rates by considering sized yarn quality. A neural network model helps to evaluate the warp breakage rates. Experimental Materials: We use actual shop floor data from mills. The fabric samples in this study are commonly used for apparel. The sample details a...
