Modeling of Induction Heating Systems Using Artificial Neural Networks


Abstract:Induction heating system has a number of inherent benefits compared to traditional heating systems. Many analytical and numerical approaches have been applied to solve the problem of induction heating. Artificial Neural Networks possess many advantages and have the ability to tackle problems that cannot be accomplished by more analytical and numerical methods. This paper involves modeling many artificial neural networks, and training them based on the results of analysis induction heating systems, by using ANSYS package, to enable them to evaluate the heat distribution inside the workpiece of any induction heating system. Also neural networks are used to specify the time and the power supply required for any desired heat distribution inside the workpiece. The neural networks are simulated by using Neural Network Toolbox in MATLAB, and the networks are trained according to supervised scaled conjugate gradient algorithm until the performance function (mean square error) reach the goal (=10-4). Artificial Neural Networks show a good success in solving the problem of induction heating through obtaining results with high accuracy and very short run time.