Predictive Control of Power Converters and Electrical Drives

11.5 Summary

This chapter presents an empirical procedure for adjusting the weighting factors of the cost function in a predictive control scheme.

Three types of cost function have been identified and a procedure for each type is presented. In this way, the first stage of weighting factor adjustment is to identify the type of cost function that will be used in the control.

At least two different figures of merit or system parameters have to be considered, depending on the application, to settle the trade-off present in the design choice of the weighting factors.

For cost functions with a primary control objective and secondary terms, the starting point is λ = 0. Then increments in the value of λ are tested until the desired behavior is obtained (branch and bound can also be used).

For cost functions with equally important terms, the cost function must be normalized, and then the weighting factor must be set to λ = 1. With this value the system will be controlled and, for fine tuning, branch and bound can be used or slight variations of λ around 1 can be tested.

Examples of the presented procedures are provided for adjusting the weighting factors of several predictive control applications.