In this paper, we study the finite-dimensional stabilization problem of the cascade of the one-dimensional transport-diffusion process with an unstable ODE plant, where it is assumed that the ODE plant is controllable and observable and the plant is connected with the transport-diffusion process through a filter. The input to the whole system is only Neumann boundary input to the transport-diffusion process, and the outputs are the data at the boundary of process domain and the output of the ODE plant. We use the previous result that the one-dimensional transport-diffusion process can be formulated as a system with A^γ-bounded output operator. In this paper, it is proved that the finite-dimensional model of the whole system is controllable and observable when the filter is a Residual Mode Filter (RMF). This fact enables us to construct a finite-dimensional stabilizing controller by using an RMF approach. The point that the output from the same RMF is used in the two parts features this paper.

抄録全体を表示

Flatness control for cold rolling mills is one of the important technologies for improving of product quality and productivity. In particular, poor flatness leads to strip tearing in the extreme case and, moreover, it significantly reduces productivity. Therefore, various flatness control system has been developed. The main actuators for flatness control are classified into two types; one is mechanical equipment such as roll bender, the other is roll coolant, which controls thermal expansion of roll. Flatness variation such as center buckle or edge wave is mainly controlled by mechanical actuator which has high response characteristics. On another front, flatness variation of local zone can be controlled by roll coolant although one's response is lower than the response of mechanical actuator. For accomplishing good flatness accuracy in cold rolling mills, it is important to improve the performance of coolant control moreover. In this paper, a new coolant control method based on flatness variation model is described. In proposed method, the state of coolant spray on or off is selected to minimize the flatness deviation by using predicted flatness variation. The effectiveness of developed system has been demonstrated by application in actual plant.

抄録全体を表示

This paper proposes a novel reduced-dimensional recurrent probabilistic neural network, and tries to classify electroencephalography (EEG) during motor images. In general, a recurrent probabilistic neural network (RPNN) is a useful tool for pattern discrimination of biological signals such as electromyograms (EMGs) and EEG due to its learning ability. However, when dealing with high dimensional data, RPNNs usually have problems of heavy computation burden and difficulty in training. To overcome these problems, the proposed RPNNs incorporates a dimension-reducing stage based on linear discriminant analysis into the network structure, and a hidden Markov model (HMM) and a Gaussian mixture model (GMM) are composed in the network structure for time-series discrimination. The proposed network is also applied to EEG discrimination using Laplacian filtering and wavelet packet transform (WPT). Discrimination experiments of EEG signals measured during calling motor images in mind were conducted with four subjects. The results showed that the proposed method can achieve relatively high discrimination performance (average discrimination rates: 84.6±5.9%), and indicated that the method has possibility to be applied for the human-machine interfaces.

抄録全体を表示

This paper proposes a new model for cooling control in the transition boiling region. The temperature dependence of the cooling process gain is expressed via a nonlinear state space form. The stability of cooling control is analyzed by applying the sum of squares technique to this model. Simulation results show that the gain of PI controller should be reduced as target temperatures become small in the transition boiling region.

抄録全体を表示

In this paper we study a method for designing a velocity-type PID controller which can assign pole and zero locations of multi-input systems. The basic idea of our synthesis is to draw a velocity-type PID control law from the state-space method which can assign poles and zeros of system in desired locations. We establish a state feedback control law in a part of input channels to attain the zero assignment and the controllability of the closed loop system. With the help of the controllability, we can deal with the pole assignment problem by means of another state feedback control law in the remaining input channels. Usefulness of the proposed method is illustrated by a numerical example.

抄録全体を表示

In this paper, we deal with a problem of encircling and navigating a target to a goal by multiple autonomous agents, where the target escapes from the agents according to a reasonable strategy. First, we derive a sufficient condition for success in both encircling and navigating the target. This result shows that the condition for both encircling and navigating is stricter than a condition for only encircling. The derived condition is described by the agents' controller gains, thus is useful to design the controllers. Second, we estimate the navigation error of the target from the goal, which might occur depending on the agents and target's parameters. Finally, the validity of the analysis results is illustrated by simulations and an experiment.

抄録全体を表示

To achieve skillful tasks with multi-fingered robot hands, many researchers have been working on sensor-based control of them. Vision sensors and tactile sensors are indispensable for the tasks, however, the correctness of the information from the vision sensors decreases as a robot hand approaches to a grasping object because of occlusion. This research aims to achieve seamless detection for reliable grasp by use of proximity sensors: correcting the positional error of the hand in vision-based approach, and contacting the fingertip in the posture for effective tactile sensing. In this paper, we propose a method for adjusting the posture of the fingertip to the surface of the object. The method applies “Net-Structure Proximity Sensor” on the fingertip, which can detect the postural error in the roll and pitch axes between the fingertip and the object surface. The experimental result shows that the postural error is corrected in the both axes even if the object dynamically rotates.

抄録全体を表示

This paper considers dynamical properties of oscillatory phenomena in gene regulatory networks with cyclic feedback. Specifically, we provide analytic estimates of oscillation profiles such as frequency, phase and amplitude based on the harmonic balance method, which is one of the frequency domain techniques to examine nonlinear oscillatory behaviors by approximating with bias and first order harmonic components. These analytic estimates allow us to find essential physical quantities that primarily determine the oscillation profiles, and they provide a guidance for synthetic oscillator design. We demonstrate the effectiveness of the proposed approach through illustrative numerical simulations.

抄録全体を表示