Virtual Realization of Temperature Measurement of PRTD Based on Radial Basis Function Neural Network

The problem of estimating the sensor´s input-output characteristic is being increasingly tackled using software techniques. Artificial neural network (ANN) method is a useful curve-fitting method. A model of PRTD (platinum resistance-temperature device) using radial basis function (RBF) neural network is presented. After the successful training of ANN using scale division data, it is then used as a neural estimator to calculate the temperature from resistance. Suitable spread parameter of the model is determined as 0.3-1.0. The absolute error is 0degC for data in specimen and 0.03~0.06degC out of specimen. The results indicate that model presented is simple in algorithm, accurate and steady. Software and hardware composition of VI system and virtual realization process of temperature measurement based on joint programming of LabVIEW and MATLAB including front panel and block diagram are given. Running results indicate that virtual realization method integrating ANN and VI is easy and inexpensive and can be used in other application fields for reference.