Burn control of an ITER-like fusion reactor using fuzzy logic

Highlights: • The burning rate of a fusion reactor is controlled with fuzzy system using a 0D model. • The fuzzy system is easy to implement and stabilizes perturbations in short times. • Performance is similar or better than other control systems such as neural networks. • System is robust with respect to random internal fluctuations. -- Abstract: An operating fusion reactor should have a fuel burning rate nearly constant in order to have a steady power exhaust. This could be achieved with a control system that monitors the produced power and modifies the fuel intake accordingly. In this work we develop such a system based on a fuzzy logic controller which adjusts the external parameters to keep the plasma temperature and density at the design values of a reactor of the characteristics of ITER. The control parameters chosen are the D-T refueling rate, the auxiliary heating power and a neutral Helium source. We use a fuzzy controller of the Mamdani type that uses a number of membership functions appropriate to produce a response to parameter deviations that minimizes the response time. The inference rules are determined in a way to provide stabilization to all initial perturbations of the temperature, density and alpha particle fraction. The dynamical response of the reactor is simulated with a zero-dimensional model that uses confinement times provided by the ITER scaling. We show how the system is feedback stabilized for a large range of parameters comprising ±25% around the nominal values. The recovery time after an initial perturbation from the steady state is within the range of one to tens of seconds depending on the type of initial perturbations applied, even with a noise term added to the energy confinement time. Furthermore the results of this Fuzzy Control System are compared with another control system based on neural networks that was previously developed.