Optimisation of combed yarn properties based on yarn number and machine jaw range using artificial neural networks

Antibacterial activity of knitted fabrics has been modelled and predicted by using two soft computing approaches, namely artificial neural network (ANN) and adaptive network-based fuzzy inference system (ANFIS). Four parameters, namely proportion of polyester–silver nanocomposite fibres in yarn, yarn count (diameter), machine gauge and type of fabric (100% polyester or 50:50 polyester–cotton), were used as input parameters for predicting antibacterial activity of knitted fabrics. For each of the input parameters, two fuzzy sets (low and high) were considered to reduce the complexity of ANFIS model. The sixteen linguistic fuzzy rules trained by ANFIS were able to explain the relationship between input parameters and antibacterial activity. A comparison between ANN and ANFIS models has also been presented. Both the models predicted the antibacterial activity of knitted fabrics with very good prediction accuracy in the training and testing data sets with coefficient of determination greater than 0.92 and mean absolute prediction error less than 5%. The robustness of the prediction results against data partitioning between training and testing sets has also been investigated. It is found that prediction accuracy of both the models was quite robust with ANFIS showing better performance with lesser number of training data.

In this study, the effects of splicing parameters, fiber and yarn properties on the tenacity and elongation of spliced yarns were investigated in detail. For this purpose, yarns from eight different cotton types, having three different counts (29.5, 19.7 and 14.8 tex) and three different twist coefficients (α tex 3653, αtex 4038, αtex 4423) were produced. Fiber properties measured using an Advanced Fiber Information System fiber tester were evaluated. Artificial neural network and response surface models were used to analyze spliced yarn tenacity and elongation as dependent variables. As independent variables, fiber properties together with the machine settings such as opening air, splicing air and splicing time, yarn twist and yarn count were chosen. As a result of the study, equations and neural network models that predict the tenacity and elongation of the spliced yarns were obtained. The obtained equations and models are statistically important and have high coefficient of multiple determination ( R2).

Thermoforming is a process where the laminated sheet is pre-heated to the desired forming temperature before being pressed and cooled between the molds to give the final formed part. Defects such as wrinkles, matrix-smear or ply-splitting could occur if the process is not optimized. Traditionally, for thermoforming of fiber-reinforced composites, engineers would either have to perform numerous physical trial and error experiments or to run a large number of high-fidelity simulations in order to determine satisfactory combinations of process parameters that would yield a defect-free part. Such methods are expensive in terms of equipment and raw material usage, mold fabrication cost and man-hours. In the last decade, there has been an ongoing trend of applying machine learning methods to engineering problems, but none for woven composite thermoforming. In this paper, two applications of artificial neural networks (ANN) are presented. The first is the use of ANN to analyze full-field contour results from simulation so as to predict the process parameters resulting in the quality of the formed product. Results show that the developed ANN can predict some input parameters reasonably well from just inspecting the images of the thermoformed laminate. The second application is to optimize the process parameters that would result in a quality part through the objectives of minimizing the maximum slip-path length and maximizing the regions of the laminate with a predesignated shear angle range. Our results show that the ANN can provide reasonable optimization of the process parameters to yield improved product quality. Overall, the results from the ANNs are encouraging when compared against experimental data. The image analysis method proposed here for machine learning is novel for composite manufacturing as it can potentially be combined with machine vision in the actual manufacturing operation to provide active feedback to ensure quality products.

Purpose – The purpose of this paper is to predict a global quality index of a ring spun yarn whose count Ne is ranging between 7.8 (76.92 tex) and 22.2 (27 tex). To fulfill this goal, a hybrid model based on artificial neural network (ANN) and fuzzy logic has been established. Fiber properties, yarn count and twist level are used as inputs to train the hybrid model and the output would be a quality index which includes the major physical properties of ring spun yarn. Design/methodology/approach – The hybrid model has been developed by means of the application of two soft computing approaches. These techniques are ANN which allows the authors to predict four important yarn properties, namely: tenacity, breaking elongation, unevenness and hairiness and fuzzy expert system which investigates spinner experience to give each combination of the four yarn properties an index ranging from 0 to 1. The prediction of the model accuracy was estimated using statistical performance criteria. These criteria are correlation coefficient, root mean square error, mean absolute error and mean relative percent error. Findings – The obtained results show that the constructed hybrid model is able to predict yarn quality from the chosen input variables with a reasonable degree of accuracy. Originality/value – Until now, there is no sufficiently information to evaluate and predict the global yarn quality from raw materials characteristics and process parameters. Therefore, this present paper’s aim is to investigate spinner experience and their understanding about both the impact of various parameters on yarn properties and the relationship between these properties and the global yarn quality to predict a quality index.

This paper demonstrates the application of two soft computing approaches namely artificial neural network (ANN) and neural-fuzzy system to forecast the unevenness of ring spun yarns. The cotton fiber properties measured by advanced fiber information system (AFIS) and yarn count have been used as inputs. The prediction accuracy of the ANN and neural-fuzzy models was compared with that of linear regression model. It was found that the prediction performance was very good for all the three models although ANN and neural-fuzzy models seem to have some edge over the linear regression model. The linguistic rules developed by the neural-fuzzy system unearth the role of input variables on the yarn unevenness.

Background3D virtual simulation prototyping software combined with computer-aided manufacturing systems are widely used and are becoming essential in the fashion industry in the earlier stages of the product development process for apparel design. These technologies streamline the garment product fitting procedures, as well as improve the supply chain environmentally, socially, and economically by eliminating large volumes of redundant samples. Buyers can easily evaluate virtual samples that are showcased with full rotation views and visual draping effects without relying on physical prototypes before confirming orders. The approved designs can be transferred to the production line immediately, which shortens the communication, development, and production lead time between suppliers and buyers. Issues of non-standardized selection on garment sizing, ease allowance, and size of 3D avatar for creating 3D garments have been addressed by many researchers. Understanding the relationship between body dimensions, ease allowance, and apparel sizes before adopting virtual garment simulation is fundamental for satisfying high customer demands in the apparel industry. However, designers find difficulties providing the appropriate garment fit for customers without fully understanding the motivation and emotions of customers’ fitting preferences in a virtual world.A statement of objective The main purpose of this study is to investigate apparel sizes for virtual fitting, particularly looking at garment ease with consideration to body dimensions and the psychographic characteristics of subjects.SignificanceThe quantitative relationship between the pattern measurements, psychological characteristics, and 3D body measurements contributes to improving virtual fit predictions for implementing mass customization in the apparel industry. This new approach and the proposed method of virtual garment fitting model prediction on garment sizes using an Artificial Neural Network (ANN) is significant in prediction accuracy. The results of this project provide sustainable value in providing an ideal communication tool between manufacturers, retailers, and consumers by offering “perfect fit” products to customers. The project will also achieve the concept of mass customization and customer orientation, and generate new size fitting data that could bring a new level of end-user satisfaction.MethodsThe study proposes to develop a virtual garment fitting prediction model using an ANN for improving virtual garment design in terms of its fitting and sizing. The project investigated apparel sizes for virtual fitting with consideration of body dimensions and psychographic characteristics of subjects on garment ease for improving the size prediction of 3D garments. We recruited 50 subjects between the ages 18-35 years old to conduct 3D body scans and a questionnaire survey for physical and psychological segmentation, as well as fitting preferences evaluation through co-design operations on virtual garment simulation using a commercial software called Optitex. Discussion of resultsThe ease preferences from subjects were significantly different from the preset values on the software. The results from the study demonstrate that ANN is effective in modeling the non-linear relationship between pattern measurements, psychological characteristics, and body measurements. The pattern parameters predicted by the ANN model were accurate. The squared correlation coefficient (R2) increased from 0.96 to 0.99 after considering different segmentations of psychographic characteristics. The ANN prediction model is proven to be an effective method for garment pattern drafting, which can achieve an individual fit and is useful for implementing the virtual fitting model.

In this study artificial neural network (ANN) models have been designed to predict the ring cotton yarn properties from the fiber properties measured on HVI (high volume instrument) system and the performance of ANN models have been compared with our previous statistical models based on regression analysis. Yarn count, twist and roving properties were selected as input variables as they give significant influence on yarn properties. In experimental part, a total of 180 cotton ring spun yarns were produced using 15 different blends. The four yarn counts and three twist multipliers were chosen within the range of Ne 20–35 and αe 3.8–4.6 respectively. After measuring yarn tenacity and breaking elongation, evaluations of data were performed by using ANN. Afterwards, sensitivity analysis results and coefficient of multiple determination (R2) values of ANN and regression models were compared. Our results show that ANN is more powerful tool than the regression models.

In this study, an artificial neural network (ANN) model is presented in order to predict the tenacity and hairiness of carded cotton yarns. Fiber measurement values generated by using a high-volume instrument (HVI) and an advanced fiber information system (AFIS) were used in the ANN model as input parameters. The radial basis function neural network (RBFNN) was used as ANN structure. The best RBFNN model was determined by analyzing the effect of epochs and the number of neurons on prediction performance. By using this ANN structure, the comparison between the performance of predicting yarn properties from HVIs and from AFISs was carried out. In the study, four different yarn counts (Ne20, Ne24, Ne30, and Ne40) for 10 different blends were applied. Each yarn count was spun at 4.34αe twist factor. In this study, the model presented a good rate of accuracy for predicting yarn tenacity and hairiness by using HVI and AFIS fiber values. The study showed that there was no significant difference between the accuracy of predicting these yarn properties from HVI fiber measurement results and those from an AFIS by using the RBF. From the results, it was noted that the performance of predicting yarn hairiness was better than that of predicting yarn tenacity. Also, this study could provide researchers with exclusive information on how to select the most appropriate ANN architecture and how to evolve the model for testing.

In this study, an artificial neural network (ANN) and a statistical model are developed to predict the unevenness of polyester/viscose blended open-end rotor spun yarns. Seven different blend ratios of polyester/viscose slivers are produced and these slivers are manufactured with four different rotor speed and four different yarn counts in rotor spinning machine. A back propagation multi layer perceptron (MLP) network and a mixture process crossed regression model (simplex lattice design) with two mixture components (polyester and viscose blend ratios) and two process variables (yarn count and rotor speed) are developed to predict the unevenness of polyester/viscose blended open-end rotor spun yarns. Both ANN and simplex lattice design have given satisfactory predictions, however, the predictions of statistical models gave more reliable results than ANN.

The dimensional shrinkage of knitted fabrics has been extensively investigated; however, existing predictive models remain limited in their ability to accurately estimate shrinkage solely from fabric properties. This study reports on the development of an Artificial Neural Network (ANN) model specifically designed to predict the dimensional shrinkage of single-jersey knitted fabrics composed of 100% cotton and polyester–elastane blends. The model integrates parameters from the knitting, pre-setting, and finishing stages, thereby providing a comprehensive framework for prediction. The training dataset was systematically compiled through controlled experimental trials on a range of knitted fabric samples, ensuring consistency and reliability of input variables. The model was trained using twenty-three input variables, including yarn count, loop shape factor, tightness factor, stitch density, course density, wale density, machine settings, and areal density. These inputs were chosen based on their known influence on shrinkage, as identified in previous literature and empirical observations. The ANN model was trained on experimental data and validated using samples not used for testing, demonstrating high prediction accuracy and a strong correlation between actual and predicted shrinkage values. The ANN was built using TensorFlow-Keras with a feed- forward backpropagation architecture, and its performance was evaluated using statistical measures, including correlation coefficients between the observed and predicted values, mean square error, mean absolute error, and mean absolute percentage error. This study demonstrates the superiority of ANN over conventional predictive models in both accuracy and scalability. Once trained, the ANN model can rapidly estimate fabric shrinkage using known input parameters, enabling proactive quality control at the production planning stage. This approach reduces reliance on physical sampling and post- compacting shrinkage testing, conserving time and material resources. The results establish ANN as a robust and practical solution to the persistent challenge of predicting shrinkage in knitted fabrics. By integrating machine learning with empirical textile knowledge, the textile industry can advance toward predictive manufacturing, improved productivity, and enhanced product performance. Furthermore, the proposed framework can be extended to incorporate parameters such as finishing and thermal treatments, and to forecast shrinkage in other knitted structures, including rib and interlock. Future research may also explore hybrid models combining ANN with fuzzy logic or genetic algorithms to strengthen predictive capability.

Fabric-evoked prickle is studied in a range of jersey knitted wool fabrics made from worsted spun yarn. The existing predictive model of relative prickliness based on earlier studies of wools with diameter characteristics is extended and can quantitatively account for changes in mean fiber length, yarn count, and fabric cover factor. For all these variables, relative prickle sensation can be predicted from the density of coarse fiber ends per unit area of fabric. It is thus possible to compare the relative importance of these variables. Within the commercial range, the mean fiber diameter of a wool remains the most important factor affecting fabric-evoked prickle.

This study aimed to model the unevenness and tenacity of ring-spun yarn in a special case in textile engineering using response surface methodology. Yarn number and front roll speed were input variables, while yarn tenacity and unevenness were response/output variables. This study showed that the response surface methodology (RSM) could predict the yarn’s tenacity and unevenness with the yarn coefficient of determination (R2) values of 0.99 and 0.98 and with the error sum of square (SS residual) values 0.00187 and 0.003215, respectively. We also found that an artificial neural network (ANN) could predict the yarn's tenacity and unevenness with the yarn coefficient of determination (R2) values of 0.51 and 0.63 and with the error sum of square (SS residual) values 1.48 and 0.856, respectively. It was concluded that the response surface methodology (RSM) and artificial neural network (ANN) could predict the yarn's tenacity and unevenness. Response surface methodology (RSM) predicts yarn characteristics better than ANN with MIMO (multiple inputs, multiple outputs) modeling. The novelty of this study is that we used RSM and ANN for the first time to obtain the tenacity and unevenness of ring-spun yarn accurately. A simpler approach was employed in this study for predicting tenacity and unevenness using RSM and ANN; however, future research holds the potential for incorporating advanced mathematical models to enhance the prediction. This research suggests that RSM and ANN can be applied to predicting the tenacity and unevenness of ring-spun yarn. The scientific application of this research is that the investigation will benefit practitioners in the textile industry to optimize yarn parameters by ring spinning machines.

The aim of this research was to identify wool characteristic of Priangan sheep (G) and its crossbred with St. Croix (H) andM. Charolais (M). The parameters observed included yield, fiber length, fiber diameter, percentage of shrink during processing,yarn production, strength and elasticity of the yarn. Eighteen rams of Priangan, HMG and MHG crossbred were used.Completely Randomized Design with 6 replications and One Way analyses were used in data analysing, except for strength andyarn elasticity, 10 replication were used. The results show that breed had no significant effect (P0.05) on yield, diameter of finefiber, and shrinking percentage during separation and carding process. In contrast, breed had significant effect on fiber length(P0.01) and on coarse fiber diameter, the shrink, strength and yarn elasticity (P0.05). In general, the wool of HMG and MHGcrossbred had better quality compared to Priangan sheep, although yarn production was higher in Priangan sheep.Key words: Wool, Priangan sheep, crossbred

An artificial neural network (ANN) is addressed for evaluating the lightning performance of high-voltage transmission lines. Several structures, learning algorithms and transfer functions were tested to produce a model with the best generalising ability. Actual input and output data, collected from operational Hellenic high-voltage transmission lines, were used in the training, validation and testing process. The method is coded in a comprehensive software program to be used by electric power utilities as a useful tool for the design of electric power systems, as an alternative to the conventional analytical methods. The aims of the paper are to describe in detail the proposed ANN method and the developed software tool and to present the results obtained by its application to operational Hellenic transmission lines of 150 kV and 400 kV. The ANN tool's results are compared with results produced from a conventional method and real records of outage rate showing a quite satisfactory agreement. overvoltages in distributions lines (13) and for lightning protection of high-voltage transmission lines (14). The calculation of the lightning performance of overhead transmission lines presents many uncertainties owing to the random nature of the lightning phenomenon and the lack of reliable data. This is why all the existing methods for the calculation of lightning performance are based on empirical and approximating equations. ANNs can be an effective alternative solution to this problem, as they can present great accuracy and, in many cases, better results using only actual line data in the lightning performance calculations. Moreover, the efficient and economic implementation of ANNs with today's computer technology makes them very attractive tools. In this paper, an ANN method is proposed to identify the lightning performance of high-voltage transmission lines. By testing several combinations of different structures, learning algorithms and transfer functions, we have developed an ANN that presented the best generalising ability among all the other combinations. Actual input and output data, collected from operational Hellenic high- voltage transmission lines w ere used in the training, validation and testing process. The developed method was coded in a comprehensive software program and applied to several operational Hellenic transmission lines of 150 kV and 400 kV so that its accuracy could be validated. The results obtained are compared with those produced using conventional methods and with real records of outage rate.

This study was carried out with aim of predicting some performance properties of fabrics with changing chenille yarn parameters. In this study, different chenille yarns were produced with parameters of pile length, yarn count and yarn type. Three different yarn types were used: polyester, acrylic and viscose. Four different yarn counts were used for each yarn type and four different pile lengths were used for each yarn count. Thus, 48 woven fabrics were obtained from 48 different yarns. The estimated properties included breaking strength in weft-warp direction and abrasion resistance, and these properties formed output data. As input data, yarn properties such as pile length, yarn count and yarn type; fabric properties such as fabric density, fabric thickness and fabric weight were used. Neural network toolbox in MATLAB was used to develop Artificial Neural Network (ANN) models. Different network structures were used to estimate three performance features, thus aiming to obtain more accurate results. Additionally, predictions were made with linear and nonlinear multiple regression models, and compared with ANN models. The R2 values ​​obtained from ANN models for breaking strength in the warp-weft direction and abrasion resistance were found to be 0.95, 0.74, 0.87, respectively, while they were found to be 0.32, 0.40, 0.41 for linear multiple regression models and 0.65, 0.27, 0.60 for nonlinear multiple regression models. The obtained ANN models were successful by a clear margin compared to statistical models.