This paper investigates a linkage control between distributed generation and power system network based on passivity characteristics. The proposed control focuses on the stored energy in inductors and capacitors for the system. It is numerically and experimentally clarified that the passivity-based control has a possibility of tracking target state with respect to disturbance input of power system network.
Recently, application of a multi agent system is expected from the viewpoint of the parallel and distributed processing of systems. Reinforcement learning attracts attention as an implementing method of the multi agent systems. However, there is a problem that the more the number of agents to deal with increases, the slower the speed of learning becomes. To solve this problem, we propose a new reinforcement learning that can learn quickly and reduce the amount of memory. It tries to increase efficiency of the learning on a tracking problem by using a method paying attention to partial states of two agents among a large number of agents.
The railway crew rostering problem aims to find an optimal assignment and cyclic sequence of crew duties to a set of rosters satisfying several labor conditions. We propose a novel decomposition approach to solve the railway crew rostering problem with the objective of fair labor condition in order to reduce the computational effort. The proposed method decomposes the original problem into two levels. The upper level master problem determines an assignment of crew duties to the set of rosters without sequencing, and the lower level subproblem generates a feasible cyclic sequence of crew duties including several resting times. Three types of effective cuts are proposed to reduce the feasible search space to tighten the gap between the solutions of the two level problems. Computational results demonstrate the effectiveness of the proposed method compared with that of the constrainted programming technique.
Adaptive control systems are designed to achieve the desired control performance when plant parameters are unknown or possibly slow-changing. In this paper, we propose an adaptive model predictive control (MPC) algorithm for a class of nonlinear input affine systems. The key idea is to combine the MPC algorithm with the adaptive Immersion and Invariance (I&I) control method. That is, MPC is used to calculate the input satisfying the assumption in the adaptive I&I control method and then the parameter update law in I&I depends on the state, estimated parameter, and input determined by the MPC algorithm. This strategy allows us to estimate the unknown parameters online and produce the control input at the same time. To modify the I&I method, we show a stability theorem for a linearly parameterized plant and then, numerical examples are given to demonstrate its effectiveness.
In this paper, a multiple-model estimation method for general non-linear systems based on particle filters is proposed. The standard particle filter for jump markov systems can easily suffer from particle degeneracy around mode change. The proposed method can cope with the degeneracy problem by controlling the number of particles in each mode without major changes of the algorithm.