In pneumatic servo systems, the adaptive pole-placement control is applied to improve the control performance. In many works, pneumatic servo systems are considered as deterministic ones. However, it is the better way to consider the plant as stochastic system in the existent pneumatic servo system, because natural oscillations of the control valves and observation noises exist. In this paper, we propose the design scheme of the minimum variance type adaptive pole-placement control for the pneumatic servo system. In this design scheme, it is based on minimizing the suitable evaluation function from the viewpoint that variance is restrained as possible against the stochastic disturbances of the plant, and control rule is led. The effectiveness of the proposed design scheme is confirmed by experiments using the existent pneumatic servo system.

View full abstract

In thermal power plants, it is an important theme to improve the control accuracy of main steam pressure and temperature etc. during load up/down. This paper focused on temperature control that is the most difficult control due to the nonlinearity and large dead time of power plant. First we constructed a simplified model of power plant that has first-order lag with dead time, and secondly a calculation method of feed forward signal is proposed from the temperature measured during load up/down. Optimal feed forward signals were decided from various load up/down rate and load change width, followed by generalization of feed foward control at any rate and load change width. Neural network was applied for this generalization. The proposed method was evaluated by a detailed simulator that represents accurately the dynamics of power plants.

View full abstract

A looper multivariable control with a disturbance observer was developed at the Hot Strip Finishing Mill of Kashima Steel Works. The new looper multivariable control consists of the following two controls which are designed so as to simplify its construction. The one is the feedforward control based on the disturbance observer which compensates for the strip tension fluctuation and looper angle fluctuation caused by the screw down position adjustments of AGC (Automatic Gauge Control). The other is the state feedback control of the strip tension and the looper angular velocity which is designed based on the pole assignment method and improves the performance of the strip tension control and the looper angle control.
The superiority of the new looper multivariable control was proved by the simulation and experimental results. The new control has been in operation since July in 1994 and improved the control performance of the strip tension and looper angle at Kashima Hot Strip Finishing Mill.

View full abstract

Spectroscopic research is performed for clarifying the pressure dependence of light emission from both excited xenon atoms and excimers produced by a silent discharge in the pressure range of 2 to 13kPa at the frequency of 60Hz between electrodes which consist of a pair of a needle and a glass-coated cylindrical-shaped metals. In the pressure range of 2 to 4kPa of xenon, two resonance lines of the wavelengths of 147 and 149nm, respectively, are observed separately. With the increase of the pressure, the intensity of those resonance lines become weak and the second continuum of the wavelength of about 172nm caused by excited dimers is eminently recognized. The first continuum does not increase with pressure compared with the second continuum. That may be caused by the loss processes such as collision relaxation. The same pressure dependence of the light intensity is also obtained for the case of krypton in the pressure range of 5 to 20kPa. At the pressure of about 10kPa of xenon, the intensity of first and second continuums becomes similar. As a result, about 30nm full width at half maximum of the emission is obtained. This shows that a wide range of vacuum-ultra-violet emission can be obtained even in one kind of gas by choosing an appropriate pressure.

View full abstract

This paper proposes an algorithm to design the modular structure in the system of software for application, and reports an example of practical use. The method represents the structure of the system by a directed graph based on the dependency of data in the system. The adjacent matrix of the graph is rearranged into hierarchical order of dependency. This re-ordered matrix gives the basis of modular design and interfaces of modules in the system. Although the outputs of this method include no executable procedures except CALL statements, but they provide reliable interface definitions of modules, which help the programmer to complete the program by residual coding in an environment of much confidence.

View full abstract

This paper presents a combined approach of rail traffic simulation using suitable models both microscopic and macroscopic. In rail traffic, the train motions do not directly depend on the preceding train motions like road traffic, but only depend on the code speed in the block section belonging at the present time. Using the history of the code speed in each block section, the simulation of each train running between stations can be performed independently. Based on this fact, we have formed a new microscopic model and proposed a combined approach applying the macroscopic model to normal train runs and the microscopic one to others. Our combined approach makes it possible to get quick responses of macroscopic simulation and detailed results of microscopic one.

View full abstract

In this paper we propose a design method of optimal deadbeat control system with state feedback. The sequence of the optimal deadbeat control signal which minimizes the quadratic performance index for the manipulated variable and deviation can be easily obtained from the matrix computation by the step respone of the system. Then, the pre-compensator and feedback gain can be derived by using sequence of the deadbeat control signal. And we also proposed a design method for not only the step-wise desired value but also the ramp-wise desired value. This design method is very simple comparing with other methods. Moreover. effectiveness of proposed method is illustratied with some numerical examples.

View full abstract

Tabu search is a novel technique for solving combinatorial optimization problems. The process in which the tabu search method seeks to transcend local optimality is based on an evaluation function which chooses the highest evaluation move in terms of objective function and tabu restrictions. This paper presents a tabu search algorithm for finding a minimum cost partition of the nodes of a directed acyclic graph into subsets of a given size, subject to the constraint that the precedence relation are satisfied. A standard tabu search approach can't find good solutions for this problem, because the problem is complex multiple partitioning problems in which number of subsets and number of nodes in each subset are unsettled. For this problem, we use appropriate data structure for this method, and develop effective neighbourhood structure and heuristics. And, we assess the effectiveness of the developed algorithm. The results show that this algorithm is effective in obtaining near-optimal solution to this problem. the running time of the procedure is proportional to the number of nodes in the graph.

View full abstract

This paper describes the results of the quantitative examination given for the first time on the ink flight peculiar to the electrically extracted ink jet method with use of dielectric ink. In the state of ink flight by this method, a liquid thread of a continuous cylindrical liquid thread is formed at the nozzle tip, and then the ink at the thread end turns into corpuscles and spreads out in mist. The author has already reported on the method for observation and on the recording for tone reproduction.
In this paper, an examination was made on the details of average and instantaneous structures of the state of ink flight, covering the wide area from the nozzle tip to the area past the electrodes, and the effect of the state of ink flight on the recording for tone reproduction was studied on the basis of the examination results. Detailed examinations were made on difference in tone, ink density distribution, relation of nozzle diameter to liquid thread diameter and to corpuscle diameter. by changing applied voltages and ink pressures. The results indicate that the ink jet which scatters in an atomized state from the liquid thread is effective in an application to a high quality recording for tone reproduction.

View full abstract

A stereoscopic cutting of stereoscopic display of 3D virtual object in a real space using a human binocular parallax is discussed. For the cutting of stereoscopically viewed image objects characterized by assumed physical properties, a scissors-type edged-tool with 2-degree of freedom is developed. Then, image-cutting force display system using the scissors-type device is synthesized by which virtual objects can be directly cut into without any particular constraint in its operation. Here, an image of sheet is given as a virtual object to be cut in a real space based on the measured physical properties. The smooth reactive force on its cutting is given to the edged-tool, through which fingers directly obtain feeling of cutting according to given physical properties of a virtual sheet material. The proposed force display system ensures a visual recognition of hand movement in operation including spatial understanding of objects with well attendance to the reality, giving a trace of the edge-point of the device by an indicator.

View full abstract

We developed a method of constituting DC reference voltage circuit which is composed of zener diode. For evaluation of the noise voltage included in the circuit, two problems have to be discussed. One is on measurement of infinitesimal voltage less than 100nV and the other is on the external noise disturbing its measurement. The former is related to the constitution of the differential amplifier utilized for amplification of its small signal. The latter is on power source noise inducing injurious voltage into the measurement circuit. This paper treats these two and gives some important suggestions for achieving accurate measurement of small signal.

View full abstract

In the field of measurement and control, there are many applications where the measurement of pipe length and pattern plays an important role. The paper proposes an automatic measurement system which makes use of the dynamic characteristics of transmitting and receiving transducers and the sonic wave propagation mechanism in the pipe system to accurately measure the length and pattern of the pipe system. To get information on the pattern and length, the system utilizes a pattern matching of the received reflected wave and the predicted one derived through the sensor's characteristics and the sonic wave propagation model.

View full abstract

It is difficult to decide the optimal neural network structure for a particular problem. We propose a solution to this problem, a new constructive learning algorithm based on division of a given learning problem. The proposed method first decomposes the original learning problem into small pieces and constructs a set of small networks which independently learn one of decomposed problems. It constructs a large network which learns the given learning problem by combining the small networks in a bottom-up manner. We demonstrate the efficiency of our learning algorithm by applying it to XOR, 3-bits parity, a non-liner function approximation, and two-spirals problem. Experimental results show that our learning algorithm can construct networks which have higher learning convergence rate and better generalization capability within less computation time than the standard back-propagation algorithms.

View full abstract