We consider ordinary differential equations (ODEs) that describe the time evolution of the concentrations of species in chemical reaction networks (CRNs). In order to analyze the convergence of solutions to the ODEs, the chemical reaction network theory has established an important theorem called Deficiency Zero Theorem (DZT). This theorem provides a sufficient condition for any solution to the ODEs to converge to an equilibrium point, based only on the graph structures of the CRNs and the algebraic properties of ODEs. In the present paper, we consider a class of non-weakly reversible chemical reaction networks, to which the DZT cannot be applied since one of the conditions, weak reversibility, is not satisfied. In order to make up for the failure of this important condition, by decomposing the network into weakly reversible sub-networks and applying the DZT to them, we show any solution to the ODEs for our class of networks with positive initial values converges to an equilibrium point on the boundary of the positive orthant.
This paper deals with a demand adjustment problem of each consumer having appliances by an aggregator based on optimal pricing in a day-ahead electricity market. In this paper, we model consumers, generators, and aggregators and design a market mechanism, where they act to maximize their own profit based on the power price and decide the electricity supply and demand. The dual decomposition is applied to the market problem to maximize the social welfare, where the proposed algorithm decides an electricity price based on the exchange of information among market participants to solve each distributed problem for improving convergence. In addition, the convergence of this proposed method is proven using Lyapunov's stability theorem. Finally, simulation results show that matching of supply and demand is achieved, while satisfying constraints on the consumption amount and the use time of each appliance, and the price determined using proposed price update algorithm converges to a certain price.
This paper deals with model predictive control (MPC) of a separated flow around a 2D circular cylinder at a low Reynolds number. The magnitude of the radial velocity on a small region of the cylinder surface is regarded as the control input. Two different objective functions are considered for each optimization step. One corresponds to the viscous dissipation function, and the other is the deviation from the ideal potential flow. In numerical simulations, flow separation and vortex shedding are suppressed by feedback control obtained by MPC for both objective functions. Moreover, a clear reduction in a drag coefficient is also observed.
In this paper, a distributed 3-D pose synchronization law is proposed for a group of networked quadrotors. The quadrotor network consisting of multiple quadrotors with dynamics and interconnection topology between them is first defined, and then, the definition of pose synchronization is provided. A novel pose synchronization law is next proposed, where linear approximation and passivation approaches are employed in the controller design. The former allows us to independently deal with the yaw, vertical, and horizontal motion dynamics of the quadrotor network, and the latter enables us to apply a passivity-based output synchronization technique. Convergence is next analyzed, and finally, the present pose synchronization law is demonstrated via simulation and an experiment to show its validity.
Control of multi-agent systems is one of the central problems in control theory. In this paper, we study the optimal monitoring (surveillance) problem over a graph. This problem is to find trajectories of multiple agents that travel each node as evenly as possible, and can be applied to several applications such as city safety management and disaster rescue. In our previous work, the finite-time optimal monitoring problem was formulated, and was reduced to a mixed integer linear programming (MILP) problem. Based on the policy of model predictive control, an optimal trajectory is generated by solving the MILP problem at each discrete time. However, the computation time for solving the MILP problem is frequently long. In this paper, to reduce the computation time, we introduce the policy of time sequence-based modeling. In the proposed method, the adjacency relation of a given graph is time varying depending on the current locations of agents. Since the unnecessary arcs are eliminated, the computation time is improved. The effectiveness of the proposed method is demonstrated by numerical examples.
Significant research on experiment-based black-box optimization using Bayesian optimization techniques is being performed because of its usefulness in a wide range of fields. Several algorithms using Bayesian optimization for optimizing environmentally adaptive control policies have been developed. This adaptivity is expected to be crucial for applications such as mobile robots. In this work, the unbiased expected improvement metric was the key to efficiently obtain the approximated optimal policy. The purpose of the metric was to remove the bias in sample points that is inevitable if ordinary metrics, such as the expected improvement, are used. This paper clarified the mechanism that causes the bias and showed that the bias should be attenuated to achieve efficient experiments. Based on the understanding of the mechanism, a simple solution was proposed to attenuate this bias. Using numerical tests, it was shown that our method effectively attenuated the bias and that this led to better optimization performance in that it often required less samples than the existing method.
This study investigates the effects of the trade-off between fuel consumption and flight time on optimal merging trajectories with allocation optimization. A merging optimization method that simultaneously optimizes arrival trajectories and the sequence of aircraft has the potential to improve future decision support systems for air traffic control. In addition, the merging optimization method is required to optimize allocation of the arrival aircraft to parallel runways because most large-scale airports have parallel runways. The merging optimization method with allocation optimization has been developed in previous studies, but only fuel consumption was minimized. Furthermore, in practice, the total flight time must be minimized, and it is possible that the trade-off between minimizing fuel consumption and flight time affects the optimal solution of the merging trajectory. Numerical simulations are performed to demonstrate the variation of the optimal merging trajectory. In particular, this study focuses on the variation of allocation of aircraft, which is a discrete factor in the merging optimization problem. The simulation results show that the allocation of aircraft can change due to the trade-off between minimizing fuel consumption and flight time. The optimality of the allocation is confirmed by comparing with the simulation results with specified allocation.
This paper deals with a gust alleviation (GA) system using gain-scheduled (GS) discrete-time preview control. The key points to improve control performance of this system are in both modeling accuracy of the linear parameter-varying (LPV) system with a small number of scheduling parameters and design of the advanced GS controller. In this study, a discrete-time LPV model of the aircraft with a small number of essential scheduling parameters is proposed through a series of approximations based on understanding of flight dynamics. A GS controller is designed with the extended linear matrix inequality (LMI) while constructing a smaller convex hull to the LPV model. The simulation result shows that the proposed control system effectively attenuates aircraft vertical acceleration in turbulence and it is robust against cruising speed changes.
This paper studies nonlinear finite-horizon optimal control problems with terminal constraints, where all nonlinear functions are rational or algebraic functions. We first extend a recursive elimination method, which decouples the Euler-Lagrange equations into sets of algebraic equations, where each set contains only the variables at the same time instant. Therefore, a candidate of an optimal feedback control law at each time instant is obtained by solving each set of algebraic equations. Next, we provide a sufficient condition such that each set of algebraic equations gives a unique local optimal feedback control law at each time instant. Illustrative and practical examples are provided to illustrate the proposed method and sufficient condition.
Profit sharing guarantees the rationality for a specific class by the rationality theorem. However it has a problem that the final result of learning is strongly influenced by the early learning phase. A reward sharing method based on safety level (RSMSL) solved this problem by introducing safety level and increasing/decreasing reward sharing rate. In this paper, we will make two improvements to the RSMSL, modification of the reward sharing function and introduction of a safety discount rate. The improved efficiency is shown in an inverted pendulum problem and a keepaway task.
This paper proposes an input design method for identification of linear-parameter-varying (LPV) systems. In particular, this paper focuses on the Bayesian estimation of LPV systems, especially the impulse responses of LPV systems at the scheduling variable of interest. The mutual information is employed as a criterion, and a concrete procedure to obtain the local optimum is given. A numerical example is shown to demonstrate the effectiveness of the proposed input design.
This paper investigates a heating, ventilation, and air-conditioning (HVAC) system in a data center equipped with a previously developed super-multipoint temperature sensing system. This system is expected to be a key technology for reducing the total power consumption of the HVAC system by controlling the inlet temperature distribution of the servers in real time. For this purpose, we present an overview of our fan-control system based on model predictive control. The main objective of this paper is to identify a dynamical model of temperature variations, in order to predict the future evolution of the distribution. However, the spatially high-density temperature data provided by the sensing system is not suited to the needed model accuracy, and the present modeling problem is differentiated from standard ones. We thus present a systematic scheme for the spatial density reduction of sensors by using spectral clustering and graph theory and associated techniques to acquire the dynamical model. Through simulation with real data, we finally show that the developed model achieves an accuracy of 0.58 degrees Celsius on average.
We investigate whether socio-psychological aspects such as empathy, grouping (swarming), and the trade-off between reactive and proactive behaviors, implemented in caribou agents improves the efficiency of the simulated evolution (via genetic programming) of their escape behavior or the effectiveness of such a behavior in the wolf-caribou predator prey pursuit problem. The latter comprises a team of inferior caribou agents attempting to escape from a single yet superior (in terms of sensory abilities, raw speed, and maximum energy) wolf agent in a simulated two-dimensional infinite toroidal world. We empirically verified the survival value of empathy in that it improves both the efficiency of evolution of escape behavior and the effectiveness of such a behavior. Also, we concluded that swarming facilitates a faster evolution of caribou agents while preserving the effectiveness of their evolved behavior. Finally, we investigated the dilemma between the reactiveness and proactiveness of the behavior of caribou agents. The experimental results suggest that the trade-off between the reactiveness and proactiveness facilitates a significant improvement of both the efficiency of evolution and the effectiveness of the evolved escape behavior of caribou agents.
An action map is one of the most fundamental options in designing a learning classifier system (LCS), which defines how LCSs cover a state action space in a problem. It still remains unclear which action map can be adequate to solve which type of problem effectively, resulting in a lack of basic design methodology of LCS in terms of the action map. This paper attempts to empirically conclude this issue with an intensive analysis comparing different action maps on LCSs. From the analysis on a benchmark classification problem, we identify a fact that an adequate action map can be determined depending on a type of problem difficulty such as class imbalance, more generally, a complexity of classification or decision boundary of problem. We also conduct an experiment on a human activity recognition task as a real world classification problem, and then confirm that a suggested adequate action map from the analysis enables an LCS to improve on the performance. Those results claim that the action map should be selected adequately in designing LCSs in order to improve their potential performance.
In this paper, we address a design method of binaural sound reproduction over loudspeakers, which is a virtual realization of three-dimensional audio. The reproduction is achieved by a model matching control to reproduce a desired sound field in a reproduced sound field. It is known that any sound field includes unstable zeros that are originated from the inherent propagation delay. Therefore, the model matching controller including the direct inversion of the reproduced sound field is inevitably unstable and cannot be implemented to actual systems. To avoid this instability, we propose a design method of a stable feedforward controller based on an H∞ model matching problem. Additional low gain specification is imposed on the controller to enhance the robustness against modeling errors of the sound field. We apply the design method to actual experimental data of a sound field and also demonstrate through a numerical experiment that the designed controller realizes binaural reproduction accurately.
In this paper, we propose a bounding method for the optimal demand-dispatch schedules in the form of time-series intervals for battery-aided electrical grids with solar photovoltaic systems (PV) when the confidence intervals of both PV output and its temporal change are known. The result corresponds to the minimal regulating capacity for generators and batteries to meet most economically power demand induced by any PV output scenario. We model a set of predicted demand scenarios as a class of convex polytope, and formulate a day-ahead economic dispatch problem as parametric quadratic programming (pQP). Then, our problem reduces to finding the interval hull of minimizers of the pQP over the convex polytope. To solve it, we prove that a minimizer of the pQP for any feasible active set of constraints can be maximized/minimized at a fixed vertex on the convex polytope, from which it can be shown that the optimal generation at any time slot reaches its maximum/minimum if power demand is maximum/minimum at any time slot, and that the optimal discharging power at each time slot does so if power demand is maximum/minimum at the same time slot and as low/high as possible at the other time slots. This means the time-series intervals of dispatch solutions can be obtained by solving the day-ahead economic dispatch problem only for a finite number of such power demand scenarios. Numerical simulations demonstrate the effectiveness of the proposed method.
This paper investigates the effect of a hearer's attitude toward a speaker's multimodal behaviors. Twenty-one university students, nine males and twelve females, participated in this experiment as speakers. Two females, aged 22 years old, performed in the role of hearers. The experiment was carried out under two conditions regarding the hearer's attitude: polite and impolite conditions. The polite hearer nodded, produced back-channel response, and gazed at the speaker, while the impolite hearer did these actions to a lesser extent. Under each condition, the speakers talked about their own experiences to the two different hearers through a within-participants design. The speech, gaze, and body motion of speakers were recorded with a video camera. In addition, the communication skills of speakers, including expressivity, sensitivity, regulation, assertiveness, and responsiveness skills, were measured after the experiment. The speakers' silent pauses, filled pauses, gaze aversion, and representational gestures were compared between the hearers' two politeness conditions. The relationship between the multimodal behaviors and the speakers' communication skills was also analyzed. The results suggest that the speakers' silent pauses, filled pauses, and gaze aversion decreased in the polite hearer condition, while representational gestures increased. From the communication-skills test, the speakers with higher expressivity, sensitivity, and regulation skills but lower assertiveness and responsiveness skills show longer silent pauses under the impolite hearer condition. Moreover, we discuss how the implications of our findings can enhance the relationship between social robots and people.