Ann based cable parameters prediction

In the cable industry, prediction of parameters of a cable to be produced has become an important issue in order to reduce the cost of manufacturing process resulting from trial and error cable in production are used and for every trail there is an associated cost and production time production. In data cable production, the cable parameters such as signal speed, Characteristic impedance, Near-end cross-talk (NEXT), and Return loss (RL) are dependent on the physical dimensions and the manufacturing parameters such as inductance, capacitance, resistance, leakage conductance of the cable per unit length and the measured" capacitance per unit length. In order to obtain the desired cable parameters, commonly trial and error methods required. Cable parameters are nonlinear functions of the physical dimensions and the manufacturing parameters. To overcome this problem, Artificial Neural Networks (ANNs) can be used due to their prediction ability. In this study, CAT6A data cable high frequency parameters are predicted using ANNs which takes the cable low frequency electrical parameters with process and manufacturing parameters as inputs and the data required to train the ANN are measured during manufacturing process of CAT 6A data cable in a cable factory. Using the experimental results, a neural network model is proposed to predict for each one of the cable parameters. The data is divided into training, test and verification data sets. The ANNs' performance are examined using the data sets. As a result, It has been found that the trained ANN can be used to predict cable parameters efficiently to manufacture CAT 6A cables.