Estimating K, C Variable Parameters of a Weighing System Using Rbf Neural Networks

Neural networks are often used as a powerful discriminating estimator for tasks in system identification. This paper describes a neural-network-based method relies on the Radial Basis Function Network (RBF network), for estimating the variable damping factor C (n) and spring constant K (n) of a weighting platform. Firstly, the RBF network learns key properties of the step response of the weighting platform and then predicts the damping factor C (n) and spring constant K (n) of other systems with different step responses before the platform settled to the steady state. In the simulation and the experimental results, with the related applied masses, the correlation rates between the actual C(n), K(n) and estimated C(n) and K(n) are presented that shows the success of this method.