A new discretization method of model equations for predictive power converter control applications based on input-state linearization

The performance of model predictive control of power electronics converters depends upon an accurate discrete-time model to properly evaluate the effect of the control action over the controlled variables. One of the most used methods to obtain a discrete-time model from the continuous-time equations is to use an Euler approximation of the derivative. However, when sampling times are increased, an Euler discretized model may incur in error in the variables. This work presents an alternative discretization method based on the input-state linearization concept of nonlinear control. An auxiliary linear model is obtained which is then discretized by an exact linear approach. The method is illustrated using the dc-dc boost type converter as a case study showing improved results a compared with the Euler approximation.