This paper deals with a seeking problem for multiple extrema by swarm robots. Extreme seeking is a task to find an extremum on a field with gradients like temperature profiles. By using swarm robots, it is expected to find multiple extrema at once. In this paper, we design a controller to spread the robots on the multiple extrema, rather than to gather them on one extremum. Our proposed method controls the variance of the robots in order to appropriately spread them out. Especially, this controller is distributed, so each robot uses only its local information.
This paper studies a data driven pole placement method deriving a state feedback gain directly from a pair of state and input measurements of a given controllable discrete-time system. In a conventional approach, a state space model should be identified in advance to apply a standard pole placement algorithm. In the proposed method, a state feedback gain and a state space model can be simultaneously obtained under an assumption.
The purpose of this paper is to propose the stochastic infectious model and to construct the optimal vaccination strategy to control the infectious disease spread. In the realistic spread of the infectious disease, environmental change and individual difference cause some kinds of random fluctuations in the model parameters. Especially, we consider the random fluctuation in the recovery rate. Moreover, focusing on the case where the number of infectives is very high, we employ the nonlinear incidence rate. We propose the stochastic infectious model with the nonlinear incidence rate. In the optimal vaccination problem, we construct the optimal vaccination system using the stochastic maximum principle (SMP). In order to solve the forward-backward stochastic differential equation in the SMP, we apply the four-step scheme and verify the efficiency of the proposed vaccination strategy through the numerical simulations.
This paper provides a new framework of control allocation of over-actuated system for static optimization of inputs via dynamical extension and potential approach. The static optimization of the control allocation is performed in real-time approximately, by varying the input according to the potential. To ensure asymptotic stability of the controlled system, we extends the backstepping method to our cases where more flexible additional potential can be used. Under the conditions on the set of the minimizer, the growth rate around the set, and the unimodality of the additional potential with some auxiliary assumptions, the extended potential function becomes a control Lyapunov function of the augmented system. The effectiveness of the proposed method is demonstrated by simulations.
Global warming and the depletion of fossil fuels encourage installing a large scale PV (photovoltaics) system. The large scale PV system may cause a issue of the voltage rise at an interconnection point due to the reverse power flow. This paper investigates the distributed management problem of PCSs (power conditioning systems) which are used to interconnect the PV system to the power grid. We consider the real-time pricing strategy of the utility, a management office of the PV systems,and each PCS determines own reference input to the reactive power flow by solving the optimization problem including the provided price. This feedback interaction of the utility and PCSs suppresses the voltage deviation. The effectiveness of the proposed distributed management methodology is evaluated through the numerical experiments, as well as, the real physical experiments.