Reliable control of high-temperature fuel cell systems using interval-based sliding mode techniques

摘要

In previous work, it has been shown that novel interval-based sliding mode techniques can be employed effectively for the design of guaranteed stabilizing control strategies for high-temperature fuel cells. These control strategies aim at both trajectory tracking during non-stationary heating phases and at compensating disturbances at desired constant operating points. The implementation of the corresponding control strategies involves three main building blocks: (i) transformation of the state equations into a non-linear controller canonical form, (ii) application of interval arithmetic techniques in real time for the computation of stabilizing system inputs despite bounded uncertainty and (iii) an online optimization for the synthesis of the input variables, namely suitable preheater mass flows and temperatures. For practical application, the transformation into the controller canonical form based on the computation of Lie derivatives can be impeded by the complexity of the symbolic representation of the state equations. Hence, techniques for algorithmic differentiation are compared in this contribution with alternative approaches based on symbolic formula manipulation. The latter one is only efficient for low-dimensional finite volume models in the description of thermodynamic system behaviour. Since the interval-based control law involves the use of time derivatives of the measured or estimated system output up to its relative degree, suitable differentiation techniques using low-pass filtered derivatives, algebraic derivative estimation and sensitivity-based approaches conclude this contribution in a numerical comparison.