A neural network identifier for electromagnetic thermotherapy systems

Purpose -This study aims to establish a dynamic process model of an electromagnetic thermotherapy system (ETS) to predict the temperature of a thermotherapy needle. Design/methodology/approach -The model is used for real-Time predicting the static and dynamic responses of temperature and can therefore provide a valuable analysis for system monitoring. Findings -The electromagnetic thermotherapy process is a nonlinear problem in which the system identification is implemented by a neural network identifier. It can simulate the input/output relationship of a real system with an excellent approximation ability to uncertain nonlinear system. A system identifier for an ETS is analyzed and selected with recurrent neural networks models to deal with various treatment processes. Originality/value -The Elman neural network (ENN) prediction model on ETS proposed in this study is an easy and feasible method. Comparing two situations of inputs with more and fewer data, both are trained to present low mean squared error, and the temperature response error appears within 15 per cent. The ENN, with the advantages of simple design and stable efficacy, is useful for establishing the temperature prediction model to ensure the security in the thermotherapy.