Characteristics of fiber optic concentration sensor, having advantages such as explosion-proof, applicability to both electrolyte and nonelectrolyte solutions, are investigated by the experiment and 2-dimensional waveguide model. Investigation shows the sensitivity is considerably increased by forming the sensor shape as the figure of 8 and further indicates the mathematical relation between the curvature of the 8-shaped sensor giving the highest sensitivity and the refractive index of the solution under measurement. This new type of sensor, made of fused silica with 1.5mm dia. and 7∼10mm radius curvature, has the resolution of the order of 0.1wt% in water solution.
In case of on-line application, it is required to correct errors automatically caused by temperature change of the solution and the contamination of the sensor. This requirement has been resolved by the development of the sampling apparatus equipped with the automatic rinsing and calibration using a standard solution. This 8-shaped sensor is successfully applied to the electric tinning plant as the tin-ion concentration meter, showing the accuracy of 2σ=0.9g/l.

抄録全体を表示

For an accurate automatic on-line measurement of ship's attitude the paper develops an intelligent sensing system which uses one servo-type accelerometer and two servo-type inclinometers appropriately located on the ship. By considering the dynamics of the servo-controlled rigid pendulums of the inclinometers, linear observation equations are derived on the rolling and pitching signals of the ship. Moreover, one accelerometer is utilized to extract the heaving signal. Through the introduction of linear dynamic models and the linear observation equations on the three signals, their on-line measurement is reduced to the state estimation of the linear dynamic systems. For unknown parameters in the dynamic models, candidates are introduced and a bank of Kalman filters are used to execute the on-line state estimation of the three signals. Furthermore, the following two methods for setting the candidates are examined to improve the accuracy of the measurement. First, all the candidates are set to be time-invariant and these values are chosen optimally from the viewpoint of the minimax criterion (method I). Second, new candidates, whose values are varied adaptively according to the changes of the unknown parameters with the time, are combined with the time-invariant candidates (method II). Both methods contribute largely to the high accuracy of the proposed on-line measurement system.

抄録全体を表示

Up to now photoelectric devices using photoconductive, photovoltaic, photoemissive and pyroelectric effects have been put to practical use. These photo detectors are devices using electrical properties of semiconductor or ceramics, while the amorphous magnetic photo detector (AMPD) is a pyro-magnetic film utilizing temperature dependence of a magnetic reluctance. The amorphous magnetic substance with an excellent magnetic property are used in electrical and electronic apparatuses, however application to a photo device has not yet been approached. The AMPD with a thickness of 20μm is made by an amorphous magnetic material with low Curie temperature and its surface is painted with a radiant flux absorbent. When the visible rays are radiated to the AMPD, the radiant flux absorbent captures the light energy and transfers thermal energy to the amorphous, which causes the reluctance change of the amorphous. The reluctance change of an AMPD can be read out as an output voltage by a Hall element. Therefore the AMPD acts as a photo-magnetic transducer. The pyro-photo detector such as an AMPD need the large reluctance change and a quick response with temperature rise. The thick film AMPD with a low Curie point agrees with this purpose and the coating of a radiant flux absorbent brings about an effective conversion of light energy. Moreover, a Hall element can read out a light signal in non-contact with the AMPD, so that an influence due to a conversion circuit does not occur. The AMPD is excepted to be used as a photo-sensor, a photo relay, a photo chopper and an exposure meter. The construction, the pyro-magnetic characteristics and the techniques of the light signal read out on the AMPD, are discussed here.

抄録全体を表示

Design of discrete-time LQ optimal regulators with prescribed closed-loop poles is considered. The desired pole assignment is accomplished by successively shifting either a real pole or a pair of complex poles. In each step the weighting matrices of the performance index is chosen such that the overall closed-loop system will be LQ optimal. Arbitrary pole assignment cannot be achieved by LQ regulators, and the region of assignable poles is clarified in this paper. A numerical example illustrates the proposed design method.

抄録全体を表示

In this paper, we study system identification in H_∞ norm using fractional representation of rational function and propose several algorithms. Model sets considered here are discrete time systems and available data are the finite series of input/output signals where initial conditions are zero. Evaluated functions used in the algorithms are coprime factors errors of a plant and its model and they are related to graph metric or gap metric. We show two methods to give the normalized coprime factorizations of the plants or the models: state space representation method using Riccati equation and Cholesky decomposition method with Toeplitz matrices. The high order coefficients of the error functions in time shift operator are unknown because data are finite. In this paper, the unknown coefficients are optimally interpolated in H_∞ norm and we give the lower bounds of the errors. Through the numerical examples we show that the series of functions which are not continuous in the sense of the graph topology can be sensitively identified by the proposed methods.

抄録全体を表示

In this paper, algebraic matrix Riccati equations are analyzed that have nonnegative-definite quadratic as well as constant terms. Necessary and sufficient conditions are established for the existence of stabilizing solution, which is not necessarily nonnegative-definite unlike the case of standard equations arising in LQG problems. Nonnegative-definite stabilizing solution is shown to exist if and only if the system is asymptotically stable and the H_∞ norm of transfer matrix be less than one. For positive-definite solution, we need additional condition that the system be controllable.
Next, necessary and sufficient conditions are established for the existence of the so-called antistabilizing solution for both general and positive-definite cases. It is shown that stabilizing solution is minimal and anti-stabilizing solution is maximal of all the real symmetric solutions. Then, assuming the existence of nonnegative-definite stabilizing solution, all the other (hence nonnegative-definite) solutions with lattice structure are derived, together with the number of nonnegative-definite solutions.
The results are established by employing consistently the so-called algebraic method based on an eigenvalue problem of a Hamiltonian matrix.

抄録全体を表示

In order to reject the steady-state tracking error, it is common to introduce integral compensators in servosystems for constant reference signals. However, if we have an exact mathematical model of the plant and there is no disturbance to the plant, the integral compensation is not needed. From this point of view, a two-degrees-of-freedom (2 DOF) servosystem has been proposed in the recent literature. The present paper considers robust stability and a transient behavior of the 2 DOF servosystem. A class of uncertainties allowed in the plant model is obtained, to which the servosystem is robustly stable for any gain of the integral compensator. This result implies that if the plant uncertainty is in the class, a high-gain integral compensation can be carried out preserving stability to achieve a high-speed tracking response. The transient behavior attained by the limit of the high-gain compensation is calculated using the singular perturbation approach.

抄録全体を表示

This paper establishes a method to calculate the Hankel singular values of systems with commensurate delays in control, using the property that the Gramian is described as an integral operator with semiseparable kernel, that is, the sum of a Volterra operator and a finite-rank one. As a result, the Hankel singular values are obtained by solving a transcendental equation.

抄録全体を表示

The purpose of this paper is to present the details of design procedure of a nonlinear regulator by Riemannian geometric approach and to apply it to the case of a double-effect evaporator. A nonlinear geometric model is proposed on a direct product space of a state vector and a control vector as well as in the previous papers by the authors. The geometric model is derived by replacing the orthogonal straight coordinate axes of a linear system with the curvilinear coordinate axes. The integral manifold of the geometric model becomes homeomorphic to that of a fictitious linear system. For the geometric model a nonlinear regulator with a performance index is designed renewedly by the procedure of optimization. The construction method of the curvilinear coordinate axes on which the nonlinear system behaves as a linear system is discussed. To apply the above regulator theory to double-effect evaporators especially to the pilot plant at the University of Alberta, a suitable nonlinear model is determined by the plant dynamics. The optimal control law is derived through the calculation of the homeomorphism. As a result it is confirmed that the regulator is effective and superior to that of the conventional control.

抄録全体を表示

Since the principle of momentum conservation holds in such an unconstrained robot as space robot, the relationship between the motion rate of manipulator hand and the angular velocity of manipulator joint is represented by the generalized Jacobian matrix. For under-water robots, unconstrained micromachines and so on, their momentums are not conservative because of the influence of the surrounding fluid and accordingly it was difficult to understand the basic characteristics for the nonconservative unconstrained robots.
By extending the concept of the impedance and the mobility of motion defined in the frequency range, the relationship between the motion rate of manipulator hand and the angular velocity of manipulator joint is derived in the frequency range for the nonconservative unconstrained robots subjected to viscous damping force. Since this relationship includes the previous generalized Jacobian matrix defined for the conservative unconstrained robots, in this study, it is defined as the dynamic generalized Jacobian matrix.
By applying the dynamic generalized Jacobian matrix to an unconstrained underwater robot representative of nonconservative unconstrained robots, its manipulatbility is considered. As a result, it is clarified that the manipulability of unconstrained underwater robot is not an ellipse and the manipulability measure is almost independent of the frequency of the manipulator joint input.

抄録全体を表示

Since thermal plants are multi-input multi-output systems, it is important to grasp the characteristic of the system for precise combustion control. Furthermore, in refuse incinerator plants (RIPs), the fuel property is unstable and minimization of exhaust emission is required. Thus, efficiency optimization from an overall standpoint with consideration of sensor and control technology is required in RIPs. Particularly, in fluidized bed incinerators (FBIs), the combustion cycle is short in comparison with stoker incinerators and combustion processes occur in multiple layers within the incinerator. For the above characteristics, dynamic character analysis is effective for grasping the characteristics of FBIs.
In this paper, we have focused on an operating FBI, and have realized a hybrid system consisting of fuzzy systems and neural networks which speculates the fuel feeding state based on measured values and combustion image processing, and operates with low CO/NOx concentration by means of air-fuel ratio control. Furthermore, we propose a tuning method for fuzzy systems which simplifies the evaluation of speculation results and the determination of control rules by utilization of an operation support system based on a numerical model.

抄録全体を表示

Recently, “Human Sensibility” is becoming increasingly important in various fields, such as design, human interface, computer graphics, computer music, and playing machines. There are a lot of factors which influence human sensibility. Color, in particular, strongly works on our mind and rouses various images.
The purpose of this work is to realize a human sensibility system for colors which extracts some images such as “pleasant-unpleasant” and “warm-cool” from R (red), G (green), and B (blue) values obtained with a TV camera as color information. The system is constructed using a neural network in which neurons are arbitrary connected and consists of two parts. One of them is a sub-network for color vision which is constructed based on its physiological and psychological knowledge. The other is another sub-network for sensibility itself. The neural network learns a human function for color sensibility from teaching data obtained by a sensory test. The outputs of the network after learning are similar to color images that we have in our mind.
The following results are proved by the analysis of the hidden units after learning.
1. The sub-network for color vision shows similar responses to those of a conventional color vision model.
2. The hidden units of the sub-network for sensibility responds selectively to “Hue” and “Tone”. The output units consolidate those selective responses and consequently the network realizes the sensibility for colors.

抄録全体を表示

Combinatorial optimization by neural networks can be roughly classified into the following two approaches. One is the application of the Hopfield type neural networks with continuous state transitions to continuous versions of combinatorial problems. The other is a direct application of the neural networks with discrete state transitions to original problems. In this paper, our attention is paid to the latter approach in order to propose the fundamentals of a new mechanism in which the artificial neurons have more advanced operational elements.
Discrete state transition rules in usual neural networks are regarded as the operational realization of the comparison between two states whose firing parts have reverse states to each other whether the firing is asynchronous or linked. The proposed transition rule realizes the comparison between more than two states corresponding to every combination of firing neurons' states at once. Concretely, the transition generates the best state among 2^N combinations of states as the outputs from the firing neurons, if the number of firing neurons is N. Such transition rules can be interpreted as a coopertive operation of plural neurons with a single input and a single output. From a different point of view, it can be a new basis to design a single advanced neuron with multi-inputs and multi-outputs which realizes the selection of a vertex with the best state in a unit hypercube composed of the single neuron's multi-dimensional state variables. This paper concentrates on a fundamental development of an advanced operation of a neuron, and limits to confirmation of its effectiveness by computer simulations for simple combinatorial problems.

抄録全体を表示

Recently, Neural Network (N.N.) modeling neuro circuits in a brain is going to be studied actively and are applied to various fields. An idea on controlling large-scale complicated systems by N.N. can be used, but it is difficult for N.N. to control the route of the network in response to input data in order to obtain better performance.
On the other hand, Petri Net is composed of state, which can express the state of the systems, and transitions, which can express various processing, and it can also control firing of the transition by tokens. So Petri Net can realize functions distribution. Functions distribution means that a specific part exists in order to realize a specific function.
In this paper, a new network model called Learning Petri Network (L.P.N.) is presented. The fundamental idea is to give Petri Net the ability of learning as N.N.. Therefore, the new network is a learning network with functions distribution that can control the route of the network in response to input data.
It is shown that L.P.N. is superior in performance to N.N. in the point of forming nonlinear discontinuous functions and identifying nonlinear systems including relay elements.

抄録全体を表示

A hybrid dynamic programming algorithm is proposed, which is based on the forward dynamic programming combined with the branch-and-bound technique. This hybrid algorithm solves nonlinear optimal control problems with complicated state and control constraints several orders faster than the conventional dynamic programming. This method overcomes the Bellman's the “curse of dimensionality”. Basically, the proposed algorithm consists of interactions similar to Darwinian evolution through the cycles of selection, multiplication through the state space quantization.
The most important idea is that this algorithm successively improves not only the optimal solution but also the accuracy of the lower bounds of the cost function which have great influence upon the number of computations as well as computation regions. To this end, the hybrid algorithm is carried out with the successive iterations under the further quantization of each state and control variable. The second idea is that this algorithm is carried out with parallel computations starting from the approximated local solutions obtained through each iteration step.
This paper assesses convergence and accuracy of the hybrid algorithm as well as the number of computations. An application of this algorithm is studied for the optimal low-thrust mission trajectory generation, in which, the Sun encounter mission is considered, and only Venus is taken as the swinged-by planet if some swingby is requested. The results show that the proposed hybrid algorithm is quite efficient for the large-scale problems with the complicated constraints such as the Sun encounter mission.

抄録全体を表示

In developing intelligent scheduling systems, it is important to determine an effective method for the acquisition of rules. This paper proposes a method for the acquisition of heuristic scheduling rules from a training set by an inductive learning and investigates its feasibility. We use ID3 learning algorithm for the rule acquisition, which generates a decision tree inductively from a training set. A training set, which plays important role in the inductive learning, is proposed to be generated by interchanging two jobs in a sample schedule in this study. The applicability of the inductive learning and the effectiveness of the generation of the training set are investigated by applying the proposed method to some scheduling problems. The effect of the size of the training set on the rule acquisition is theoretically and numerically analyzed from the entropy view point. Firstly, we apply the method to one machine scheduling problems, of which simple optimum rule is known, and show a possibility of the acquisition of rules by our method. As a result, it is shown that the obtained rules are equivalent to the optimum rules. Secondly, by applying the method to one machine problems of which simple optimum rule is not known and the flow-shop problems, it is shown that effective rules can be acquired. Lastly, we demonstrate the effectiveness of a two phase method to obtain a sample schedule for the second phase acquisition by applying the rules obtained by the first phase acquisition.

抄録全体を表示

In this paper, impedance characteristics of human arm during isometric muscle contraction are measured experimentally, and relationships between hand impedance and various muscle contraction levels are analyzed. Firstly, while a subject regulates a muscle contraction level or amplitude and direction of hand force to an instructed value, small external disturbances to his hand are applied by a manipulandum. Time changes of the hand displacements and forces caused by the disturbance are measured, and the hand impedance are estimated using a second-order linear model. During the experiments, EMG signals of major muscles acing on shoulder and/or elbow joints are measured in order to estimate a muscle contraction level of the subject. The experimental results in different subjects are summarized as follows: 1) coactivation of agonist and antagonist muscles increases the area of stiffness and viscosity ellipses, 2) spatial characteristics of hand stiffness are changed depending on the amplitude and the direction of the hand force, 3) not only stiffness but also virtual trajectory of the subject's hand change with hand force, 4) it might be possible to predict the arm impedance form EMG signals with sufficient accuracy.

抄録全体を表示