This paper proposes a hierarchical system representation of large-scale multi-cellular networks coupled by the quorum-sensing mechanism and present a control theoretic approach to systematically analyzing the local stability and instability of an equilibrium state. In particular, we are concerned with the dynamics of coupled Repressilators and analytically derive conditions for the local instability of a homogeneous equilibrium. It is first shown that the dynamics of the quorum-sensing network can be formulated as a dynamical multi-agent system with a hierarchical structure having low rank interconnections. This structural property then allows us to decompose the high-order dynamics of the coupled Repressilator network into two low-order systems. Thus, the proposed approach significantly reduces the complexity of the local stability analysis and leads to the analytic necessary and sufficient local stability conditions of a homogeneous equilibrium point. The analytic conditions reveal the reaction parameters that essentially affect the existence of oscillatory dynamics in the coupled Repressilator network.
A reduction method of residual positioning error of balancing balls on an auto-balancer mechanism is proposed. Rotation speed profile control method improves the performance of auto-balancer by providing a constant speed period at a critical speed in the middle of motor acceleration. In order to confirm its effect, equations of motion of the auto-balancer mechanism including balance balls are modeled based on non-stationary vibration model. Time-variant dynamics of the auto-balancer in accelerating the motor is simulated by using MATLAB and the balance balls are confirmed to move toward a balanced position. Furthermore, an experimental study by this method is performed on an optical disc drive with the auto-balancer and vibration amplitude suppression is verified.
Temperature measurements of liquid pig iron coming out of a blast furnace are essential for estimating the thermal condition in the hearth. The temperature is generally monitored by a disposable thermocouple on a runner. This paper proposes radiometric temperature measurement that targets an iron-slag-mixed stream in front of a taphole. It enables non-contact continuous thermometry. Liquid iron and slag are spatially separated on a thermal image obtained by a CCD camera with a high-speed electronic shutter. Molten iron temperature is calculated from the iron radiance detected automatically by means of histogram processing of the thermal image. Regarding optically semitransparent molten slag, its radiance varies as a function of the thickness. Slag temperature is determined from the highest radiance, presuming that the emittance of sufficiently thick slag converges on a constant. The authors performed an on-site test at a blast furnace in ordinary operation. It was confirmed that both iron and slag temperatures were continuously monitored. The temperature data obtained from the test showed findings such as unsteady temperature difference between the two liquids and discontinuous temperature fluctuation. This imaging thermometry technique is expected to be used in a new sensor for blast furnace operation.
This paper investigates model predictive control (MPC) of large-scale systems with both continuous-valued and discrete-valued control inputs. In the authors' previously proposed method, the notion of a virtual control input, which is obtained by relaxing a discrete-valued control input to a continuous decision variable, was introduced. In online computation, first, a continuous-valued control input and a virtual control input are calculated. Next, using the virtual control input obtained, only a discrete-valued control input at the current time is calculated. In this paper, as remaining issues, quantization errors, stabilization, and an application to air-conditioning systems are discussed. The results in this paper enhance the effectiveness of our proposed method in large-scale MPC.
This paper presents a novel mesh refinement technique in pseudospectral methods for numerically solving nonlinear optimal control problems. Pseudospectral methods are known to suffer from high computational cost and deterioration of convergence speed when the solution is not sufficiently smooth, which are matters of concern when solving realistic optimal control problems arising in engineering applications. The proposed algorithm automatically and iteratively provides a discretization mesh which overcomes these drawbacks. Its main feature is detection of difficulties in the polynomial approximation of the solution and isolation of these difficulties in smaller segments. This segment decomposition improves numerical efficiency and stability while making the best use of the potentially exponential convergence property of pseudospectral methods. Validity and efficiency of the proposed technique are demonstrated through solving two example problems.
This paper proposes a concept of a robotic fire-fighting system which performs fire extinguishment by a swarm of aerial extinguishers with inert gas capsules. This paper also develops a prototype of an aerial extinguisher and its control system. It is demonstrated that the extinguisher achieves flame extinguishment. This confirms a fire extinguishment ability of the proposed fire-fighting system and a mobility of inert gas capsules.
In recent years, the authors have proposed an effective metaheuristic method for continuous optimization problems based on an analogy of spiral phenomena in nature. This method is called Spiral Optimization (SPO). SPO has two setting parameters: the convergence rate and the rotation rate. Depending on computational and/or problem conditions, the values of these parameters affect search performance. However, effective setting methods for these parameters without trial and error, including analyses of search dynamics needed for such methods, have not yet been studied. This paper analyzes the stability of the dynamic equilibrium point of the SPO model and proposes a parameter-setting method for the convergence rate r based on results from stability analysis. The effectiveness of the proposed method is confirmed through simulation for some benchmark functions and comparison with other representative metaheuristic methods.
This paper addresses movement and observation planning for mobile robots under position uncertainty. A sequence of actions that minimizes the total time needed for reaching a destination while guaranteeing that the collision probability is less than a prescribed threshold is planned. The problem is formulated as a path planning problem on a roadmap with additional constraints on covariance matrices expressing position uncertainty, for which a novel branch-and-bound based solution is presented. Moreover, a heuristic technique for creating a roadmap based on a new criterion that encapsulates both collision safety and localization ability is proposed. Simulation study is performed to evaluate computational complexity and relations between some characteristic parameters and obtained solutions.