Abstract:
Parallel multi-cell converters using inter-cell transformers are real multi-input multi-output systems, making their control challenging and possibly requiring increased embedded computing power in the control architecture. The challenge of the current study is to design a control algorithm as simple as possible, in terms of settings and implementation, while meeting standard specification. The state-space representation of multi-cell converters permits to define a full state feedback. Such Multi-Input Multi-Output (MIMO) systems have numerous tuning parameters which enable various ways to tackle the control specifications. Among the specific approaches, both total decoupling and optimal control based on quadratic cost and objective functions are addressed thoroughly and consistently. The studied case is a 3-cell parallel converter for which various settings of the state feedback are considered and analyzed by simulation Linear-quadratic regulator design reveals the best compromise between variables tracking precision and robustness towards system parameters and load variation. Furthermore it is easy to implement utilizing few non-zero setting coefficients. Specifically the feedback gain matrix associated to the integral terms is almost diagonal: this natural decoupling makes it extremely simple to efficiently implement an anti-windup algorithm. This is an important result since until now, the standard solution is mostly based on decoupling strategies. Among other drawbacks, this latter approach proves to be much more sensitive to parameter uncertainties.
