High-dimensional data require a lot of computational work for processing. In order to efficiently and accurately obtain the results of measurement from high-dimensional data, significant features should be extracted before processing. Here we present a feature extraction method for significance weighted supervised classification. Conventional methods of extracting features consider only the average classification accuracy. In contrast with this, we present a new method of feature extraction which is based on a significance weighted criterion. The purpose is to extract features which separate particular pair of classes and give high classification accuracy for important classes. In the first step, assuming that all the classes have the same within-class covariance matrices and normally distributed, all the data are reduced by principal component analysis. Most of the information can be expressed in low dimensional space because many of the dimensions of high-dimensional data are highly correlated. In the next step, feature vectors are determined in the space of reduced data. Each feature is extracted successively by selecting the feature vectors which separate the hardest-to-separate pair of classes among the important pairs of classes to be separated. A feature vector which separates two classes is set according to the Fisher's linear discriminant function. The method is applied to about 500 dimensional hyperspectral data which are required to be classified into five categories. The feature extraction technique, together with the results of numerical simulations which confirm the validity of this approach are presented.
This paper describes about the development of a touch sensor, which has a mono-structure, but has two kinds of functions for discrimination of material properties. The proposed sensor is used to discriminate materials such as iron, aluminium, copper, bronze, rubber, wood, acrylic resin, styrene foam. The sensor consists of two spiral metallic lines and has four terminals. When the two lines are used as two coils, the sensor has the property of inductance. Otherwise, when the two lines are used as two parallel electrodes, the sensor has the property of capacitance. The experiments show good results to discriminate each sample of materials and are expected as a step to develop an artificial skin tactile sensing system in future.
Measurements of flow velocities using optical techniques have advantages to measure flow velocity in a microscopic region without disturbing it. However the improvement of the spatial resolution in the direction of the optical axis (the depth of the measuring region) is difficult for even the optical techniques. In this study a new velocimeter with high spatial resolution is proposed. The velocimeter is based on a laser two focus velocimeter (L 2 F) and its optical system is constructed with a confocal system to improve the spatial resolution in the direction of the optical axis. The advantage of the confocal optical system is that the light from the only part within the focal region is detected strongly. The depth of the measuring region of the proposed L 2 F is about 8.2μm. The reliability of the proposed L 2 F is confirmed by velocity measurements of moving a tungsten wire. The results of the experiments show that the proposed L 2 F with the confocal optical system achieves velocity measurements with high spatial resolution and it is extremely effective instrument to measure the flow velocities in microscopic regions.
The measurement of angular velocity and angular acceleration in sports biomechanics is usually based on cinematography. But this method dose not have good accuracy. Authors previously developed angular velocity and angular acceleration measurement system, which used 12 accelerometers. But this system is tedious, beacause of mounting the large number of accelerometers. This paper proposes a measurement system using 9 accelerometers in sports, which accurately measures the rotation motion. This measurement system uses the extended Kalman filter. The accuracy of this method was examined by simulation and experiments.
There is no work on odor or organic gas visualization using a fiber-optic sensor array, although a semiconductor gas sensor array has been used for odor visualization. This paper proposes a new method of visualizing odor or organic gas distribution changes with time, which method uses a fiber-optic sensor array with dye coatings. A 32-channel fiber-optic fiber sensor array was made using a plastic fiber and a methylene blue dye/silicone coating for visualizing methanol gas distribution. The optical absorption change signals from the 32-channel sensor array were detected by a 32-channel photodiode array and stored into computer via 32-channel I/V converter, amplifier, and A/D converter. A two-dimensional methanol gas concentration distribution and its changes with time were demonstrated on computer display by using the system.
It becomes necessary for dental diagnosis and treatment of patients having complaints with their temporomandibular joints to precisely measure motions of their lower jaws. Although such devices as can measure one-dimensional motions of a magnet attached to an examinee's incisor or 3-dimensional mandibular movements by means of mechanical displacement transducers have already been introduced into actual service, there has been great demand for more practical and reliable ones. The expected specifications of the device are as follows; (1) high accuracy and reliability, (2) 6 degree-of-freedom measurement covering wide range, (3) light and handy with little physical burden, (4) virtually real-time data processing. The present paper intends to develop a new mandibular examination device using PSD which can meet these demands as much as possible. The characteristic relation between the output voltages of a PSD sensor and the spatial position of a target LED, which relation basically dominates the accuracy of measurements, has been obtained on the basis of geometrical optics, proving through calibration that detecting errors remain less than 0.1mm with LED positions. The device has been constructed of two light duralumin arms which support three PSD detectors and three target LED's, respectively, and are to be fastened to the examinee's upper and lower dental arches. Test measurements of plane and rotational motions of the LED arm created by an automatic pulse stage have revealed the accuracy and usefulness of the device; average deviations between measured and programed trajectories were within 0.2mm. Applications of the device to two subject examinees, one is normal and the other has a temporomandibular joint disorder, have been successful, showing its effectiveness as a examination device. With the electronic system and PC used, it took 30 seconds to conduct one whole run for an examinee, from the start to the final acquisition of graphical data display on CRT.
We will propose improved method to measure the macro shape of long object moving with fluctuations. There already exists similar method, but it has fatal defects. The unsuitable data processing of existing method often leads the macro shape of obtained result far from the reality. So, we have established improved method to remove above mentioned defects of existing method, and successfully improved the accuracy of measured results. The system of improved method as well as existing one consists of three distance sensors arranged in a row not necessarily at regular intervals along the transfer line. As the object proceeds, the distance values to the moving object are acquired simultaneously by each sensor. At every occasion of data acquisition, we will obtain set of three equations consist of acquired data, angular and to & fro elements of fluctuations as unknowns and also unknown coefficients of lower powered function to represent the macro shape of long moving object. After completing data acquisition over the whole body of moving object, all the sets of equations are processed by the computer using Least Squares Method to obtain the macro shape, in other words, to decide the values of coefficients of lower powered function. The main feature of our proposal is to improve the way to solve the simultaneous equations and successfully to remove above mentioned defects of existing method. The favorable characteristics of improved method are that it can not only cancell the effects of fluctuations but also provide us with undeformed macro shape of obtained result.
The status of combustion of the pulverized coal may be influenced by the shape, especially, of its surface area. Thus, the measuring of the surface area is one of the most important tasks. Surface area of not a single particle but particles has been measured by gas adsorption method and permeation method. The authors have proposed an image processing technique with microscope as an on-line measurement particle size distribution and particle shape coefficient. In this paper, we present a new method to measure a surface area of pulverized coal with an imageprocessing technique using two optical microscopes. The measurement errors obtained by simulations using regular shape particles were less than 10% and the accuracy was discussed on the basis of simulated results. Surface areas of two kinds of pulverized coal were measured by the method and we showed the measuring results.
Many measurement methods of velocity vector distributions by image processing have been developed and applied to various flows. Most of them, however, have a serious problem. The estimated velocity vectors in a flow field include some erroneous ones because of mismatching a tracer image pattern to another one. Accordingly, it is necessary to determine the erroneous vectors which exist in the estimated velocity vector distribution. This paper presents a new algorithm for determining the erroneous vectors using Genetic Algorithms, which are search algorithms based on the mechanics of natural selection and natural genetics. With genetic algorithms, the first step of the optimization process is to code the estimated distribution as a binary (0, 1) matrix, where “1” and “0” should correspond to right and erroneous vectors, respectively. Second, many individuals with binary matrices, which have random and different (0, 1) combinations, are prepared as an initial population. Third, genetic operations such as “Reproduction”, “Crossover”, and “Mutation”, are performed to the prepared candidates and are repeated step by step according to a fitness function which evaluates similarity between neighboring velocity vectors. Finally, an optimal solution of the optimization problem is obtained from a binary (0, 1) matrix with the highest fitness after thousands of generations. The Genetic Algorithms are applied to two velocity vector distributions obtained by computer simulation and image processing respectively, which includes erroneous vectors. After thousands of generations, most of the erroneous vectors are determined and eliminated. As a result, it is proved that this new algorithms can distinguish erroneous vectors from right ones by checking over the entire measured flow field even if the density of erroneous vectors in a flow field is high.
A new method for grouping of relevant and equivalent inputs of a logical circuit is proposed by making use of input-output correlation function. Pseudorandom M-sequences are applied to a logical circuit and their outputs are observed. The crosscorrelation function between the input and output gives us the information on the logical circuit whether it is faulty or not. The authors show that it is possible to classify the faults by use of this grouping method. The authors also show an example of fault estimation by this method, indicating further application of this method.
In this paper, attention is focused on the quantitative understanding of the role of sound intensity which is considered to be a very important cue of the sound localization with distance. Experiments were conducted by using virtual environment to clarify the relation between sound intensity and apparent distance from the sound source. The results are the followings: (1) Replacing the experiments by M. B. Gardner to the virtual environment, the experimental results obtained on virtual environment were equal to those of real world. This fact ensures that the virtual environment can give the same results as those from real environment. The experimental condition of no other cues except sound intensity confirms the conclusion by M. B. Gardner that the sound intensity is a very important cue of apparent distance from the sound source. (2) The relation between sound intensity and apparent source distance can be expressed as the form of exponential function. Namely, it becomes a straight line in logarithmic graph. (3) The coefficient of the proportion of the straight line in logarithmic graph is restricted by both of the range of sound source distance and the range of sound intensity. The coefficients without any restriction are larger than the results obtained by M. B. Gardner. (4) The results obtained from the experiments using method of adjustment was equal to those obtained from the experiments using forced-choice method with ranges wide enough. These results show the coefficients obtained are independent from any experimental conditions. The obtained results above will be quite useful to design and evaluate the system using acoustic display.
Buoy tracking system with the GPS navigator and MCA transceiver was developed to measure the water movement in Lake Biwa, Japan. The accuracy of positioning by GPS is about 100m which is sufficient for the measurement of gyres in the lake. The MCA (multicannel access) radio system was developed for making good use of limited radio frequencies, and 15 radio communication channels are shared by many users. It enables data transfer of buoy locations, which are obtained in real time at the base station. Buoy location is always known, and buoy recovery is easy. A typical example of a buoy track observed in Lake Biwa is presented.
Parametric absolute stability is the concept of stability which deals with feasibility and stability of equilibrium states of Lur'e systems with parametric linear parts and sectorial bounded nonlinearities. A Popov-type sufficient condition for parametric absolute stability has been obtained for single-variable Lur'e systems. For multivariable Lur'e systems, a condition in the state space, which is described by linear matrix inequalities (LMIs), has been derived. For polytopic Lur'e systems, the LMIs are easily solved by a computational tool. However, if the linear part is nonlinear with respect to parameters, there exists no useful tool solving the LMIs. In this paper, the Popov-type condition of parametric absolute stability is extended to multivariable Lur'e systems. An example is presented, in which the condition is tested by applying the polygon interval arithmetic (PIA).
Stabilization of a linear time-invariant single-input single-output plant with a variable operating condition is considered. The plant is represented as an interpolation of two nominal models described by proper stable factorizations of their transfer functions. An interpolation of stabilizing controllers for the nominal plant models is applied to the interpolated plant. The stabilization problem is reduced to a Nevanlinna-Pick interpolation problem, and the necessary and sufficient condition is derived for the existence of a stabilizing controller. It is shown that the class of systems stabilizable by interpolated controllers is larger than that stabilizable by fixed controllers.
In this paper, the finite pole assignment problem for distributed parameter systems is considered. This is formulated as the problem to assign a finite number of eigenvalues of distributed parameter systems, which have infinite number of discrete spectrum, to predetermined positions. The necessary and sufficient condition for the ability of the finite pole assignment for the non-selfadjoint operator case is shown. An algorithm to calculate feedback gains assigning poles approximately is developed by using the Galerkin approximate method. The convergence of the resultant approximating control is studied.
A new method of designing an observer-based adaptive controller is proposed for a class of singleinput single-output nonlinear systems, which have no unknown parameters in the term multiplied by the input. This method doesn't require the measurement of the states. Furthermore, the systems are only required to be transformable into the adaptive observer form, not into the output-feedback normal form. Consequently, this method can be applied to the systems which the existing methods can't be applied to.
This paper considers the H∞ filtering problem with frequency constraints on the unit circle of the complex plane. If the system is subject to step or periodic disturbances, the state estimates may be degraded due to the biases or periodic fluctuations. In order to remove these undesirable effects, we impose constraints such that the transfer matrices from the disturbances to the error must be zero at certain frequency points on the unit circle. Based on the Nevanlinna-Pick interpolation technique, we develop a method for adjusting the free parameter of an H∞ filter so that the constraints on the unit circle are satisfied. We also show that the resulting H∞ filter is a linear function.observer for the augmented system including the disturbance model. A numerical example is included to demonstrate the applicability of the present design method.
In this paper, a new design scheme of the simple adaptive control (SAC) system based on a frequency domain analysis is considered. The SAC is applicable only to almost strictly positive real (ASPR) plants. However, we can construct an ASPR augmented plant, which is virtually regarded as a new controlled plant, by implementing a parallel feedforward compensator (PFC) or pre-compensator on the plant. The design scheme of such compensators which make the non-ASPR plant to be virtually ASPR is proposed by taking plant uncertainties into consideration. As a result, the applicable class of the SAC can be expanded.
Here the controller design method for a kind of nonminimum phase systems which sometimes appear in controller design problems for automobiles, aircrafts, or large scale systems is considered. In these cases a unstable zero which locates near the origin may limit the performance of the controlled system. Especially it is concerned that the proposed controller design method is such a one that the available closed loop performance is easy to see for a designer when he knows the location of the unstable zero. In the proposed method, the performance, disturbance attenuation, robustness and the location of the closed loop poles are taken account through loop-transfer function gain shaping with the clssical stability margin concept. To do so, the framework of H∞ mixed sensitivity problem is used and how to choose the frequency weights is proposed. This method can be applied to servo system design or plants having poles on the imaginary axis. And also undesirable pole-zero cancellation coming from the mixed sensitibity problem is prevented by proposed frequency weights choice.
In this paper, a model following control problem (NMFCP, for short) for nonlinear MIMO systems is studied. Isurugi and Shima has already given the solution of this problem under some regularity assumptions. The result can be stated as follows: NMFCP is solvable iff the relative order of the plant is not greater than that of the model. We consider the case where this condition is not satisfied. In this case, although any feedback law cannot cansel the influence of the inputs of the model on the error, the high-gain feedback based on the singular perturbation theory is available. It is shown that, by using the high-gain feedback with some small positive parameter ε, the error will remain o(ε). Finally, the efficiency of the proposed control law is illustrated by a numerical simulation.
In the design problem of linear discrete-time control systems with input and state constraints, a feedback gain is admissible if all the constraints are satisfied. This paper presents the design method for obtaining an optimal linear admissible control law using a criterion for admissible gain when the set of initial states is given by a convex polyhedron. This problem has attracted special interest recently. The positively invariant condition, which assures that the trajectories of control system from the set of initial states stay within the set, is often used to design the control law. In fact, this condition is too conservative, so that it can cover only a narrow region of admissible gains, and the resulting control system is in general not optimal in the range of admissible linear laws. A method for computing the maximum value of constrained variables, α(f), is derived from the study in the dual space; this gives the necessary and sufficient condition for admissible gain and is more efficient than the method proposed by Kiendl because it requires less trajectory computations. The design problem is formulated as the following programming problem: find an admissible gain minimizing a given performance index. As for the performance index, either a weighted quadratic sum of the states or a size of the state at a specific time is considered. It is also shown that the solvability of the problem can be examined by solving the minimization problem of α(f). Gain gradients on the function α(f) and the performance indices are derived, which are used in the optimization algorithms. The nonconvex property of these problems is pointed out and for each problem an algorithm finding a suboptimal gain is proposed. A 2nd-order control system design problem with constraints is treated as a numerical example, and satisfactory results are obtained, which prove the effectiveness of the design algorithm.
An adaptive boundary control problem for a stochastic heat diffusion equation is studied. The considered system contains the unkown potential coefficient which is a function of a spatial variable. The estimation algorithm for the unknown potential coefficient is proposed by using the stochastic approximation technique. After showing the strong consistency of the estimated parameter, the cost for the adaptive control scheme presented here is shown to converge to the optimal ergodic cost. Finally some numerical examples are shown.
This paper proposes a model of nonlinear dynamical systems which consists of a bank of orthogonal filters and a three-layer artificial neural network. The relation of this model and the parametrized Volterra series model is presented. After executing an ordinary learning process on the artificial neural network, the Volterra kernels are directly calculated in terms of the connection weights of the network. One advantage of the present method is that kernels of arbitrary orders can be obtained independently no matter what the order of the Volterra series model is. Another advantage is, by examining the obtained connection weights, it is possible to tell whether the identified system has a special structure referred to a Wiener model which consists of a dynamic linear subsystem followed by a static nonlinear subsystem. The theoretical results are tested by numerical simulations.
In view of trajectory formation in human arm movements, various objective functions are proposed such as jerk for a hand, torque-change of joints and so on. However, these are not determined from physiological characteristics of a human arm, and it is difficult to explain directly the necessity why humans select such trajectories. In this study, an objective function which is based on expended energy in muscles and joints of an arm is proposed. Experiments have been conducted for movements of a human arm in a vertical plane passing through the shoulder so that movements are influenced by gravity. Then, measured trajectories are compared with those formulated with the objective function. The results show good agreements between measured and formulated trajectories. It is also verified that different types of trajectories are selected according to humans and the proposed objective function retains applicability to formulate the corresponding trajectories which vary with humans. Finally, trajectories selected by humans are analyzed with the formulated trajectories, and the physiological and dynamic characteristics of human arm movements are discussed.
In this paper, the authors propose a new identification method for the discrete-time impulse response model of a linear system from sampled input-output data using multiresolution analysis theory, especially in the case of impulse response with locally rapidly changing components. The continuous-time impulse response of the system under study which is viewed as a L2(R) function of time, is approximated by scaling and wavelet functions which are shifted and dilated based on the multiresolution analysis theory. Hence the system under study can be viewed as weighted summation of a group of subsystems in which the shifted and dilated scaling functions and wavelet functions are interpretated as their impulse responses respectively. Then the genetic algorithm and AIC are introduced to select significant subsystems such that only moderate parameters are required to be estimated in contrast to the conventional method. It is shown that the proposed method yields accurate estimate of the impulse response with locally rapidly changing components, even in the ill-posed cases of band-limited input, fast sampling rate and significant measurement noise.
In this paper, systems with bounded disturbances and unknown relative degree not exceeding three are considered, and we propose a design method of model reference adaptive control systems. The objective considered here is to design the control system so that the norm of the output error, between the output of a controlled object and that of a reference model, converges to any small region in the neighborhood of the origin. In order to achieve the objective, the unknown parameters and the unknown bounded disturbance are estimated by using the σ modification method, and the control system is constructed by using the estimated parameters. In this control system, we derive the relation between the output error and the design parameters. From this relation, it will be shown that the upper bound of the norm of the ultimate output error can be reduced to any small value by increasing the value of only one design parameter. Moreover, if the prior information about the upper bound of the norm of the unknown parameters is known, it will be shown that, when no disturbance is present, we can construct the control system by using the prior information in which the convergence of the output error to zero is guaranteed. Finally, computer simulation results are presented to illustrate the effectiveness of the proposed method.
This paper presents an optimal control of liquid sloshing caused by transferring a container from the view point of transferring time and maximum change of liquid level. A two-dimensional rectangular type container partially filled with water and a linear transferring track are considered. In order to analyze the higher-order sloshing modes which may occur in transferring the container, computer simulation has been conducted by using the boundary element method (BEM). The basic equations of sloshing are Laplace equation in terms of velocity potential and pressure equation on free surface. Control design is done based on the modal model derived from the frequency response obtained by BEM simulation. An optimal servo system of LQI control with Kalman filter is applied in oder to find an optimal transportation control for the given performance specification, where an optimal weighting matrix in the linear quadratic problem may be determined by Carroll's created-response surface method together with simplex method. Influence of a spillover by truncating the higher-order modes is studied on the present feedback control system through both BEM simulations and experiments. Several control experiments have been done to demonstrate the usefulness of the proposed control system.
The purpose of this paper is to present a method of the control design for a nonlinear process to swing up the inverted pendulum which stay initially in the natural pendent position and to stabilize the pendulum at the inverted position. Firstly, to swing up the inverted pendulum, optimal trajectory is derived by bang-bang control such as makes a cart move into right and left direction in turn near the natural frequency of the pendulum. Secondly, a nonlinear differential equation is transformed into a linear time-varying system around the optimal trajectory (nominal state), and then a linear time-varying optimal regulator is applied to make the pendulum trajectory exactly follow up the nominal state, and then to stabilize it at the inverted point. After a pendulum is inverted, a time-invariant optimal regulator with higher control gain than time-varying one is used in order to reduce the off-set of a cart position and to stabilize the pendulum for disturbances. Further, Kalman filter is applied to eliminate the small vibration of pendulum. Finally, it is demonstrated that the successful probability for swinging up and stabilizing has been almost 100% by the control experiments. The presented method improves the result of the previous research and shows to produce a reliable control system.
A new trajectory planning method using tactile feedback is investigated for the manipulation of an object by a multifingered hand where the fingertips and the object make rolling contact. The proposed control algorithm is capable of determining the motion of each finger so that the object is manipulated along the desired trajectory according to the tactile feedback at the fingertip of the hand. The advantage of the algorithm developed here is that geometrical information about the object to be manipulated is not necessary and therefore can be applied to the manipulation of an unknown object. First, the mathematical model concerning the kinematics of the manipulation with rolling contact at the fingertip is formulated. Next, assuming that the fingertips are equipped with a tactile sensor for detecting the location of the contact, the trajectory planning algorithm for the fingertips is derived. Finally, the previously developed finger-shaped tactile sensor using an optical waveguide is installed on the two-fingered hand, and the superiority of the proposed manipulation algorithm is confirmed through computer simulations and experiments.
In this paper we treat a three-inputs/three-outputs magnetic levitation system with three electromagnets and a Y shape iron plate as a levitated vehicle without mechanical contact. In general, the magnetic levitation system possesses nonlinearity, natural frequency, and parameter variations inherently, so that we design a decoupled attitude control system in order to compensate nonlinearity and sensor noise. In this paper we compensate the nonlinearity by using a method of exact linearization and design the control system by using ILQ design method. This design method can achieve decoupling and specified time responses, which is important in this multivariable control system, and we can reduce sensor noise by choosing the design parameters appropriately. In this paper we first derive two linear models by using a method of exact linearization as well as a usual method of linearization around an equilibrium point. Next we design control systems by using ILQ design method and show the effectiveness of both the design method and two linearization methods by simulations and experiments. Finally we show that with exact linearization method, 1) the derivation of the model is easier, 2) some parameter variations can be treated in the framework of quadratic stabilization problem or μ synthesis, 3) the selection of the design parameters does not depend on operating points, and 4) ILQ design method can eliminate some parameter variations by tuning certain design parameters.
The option pricing theory has, recently, been studied by many scholars in the fields of finance, statistics, and control theory. The evaluation of option price has close relation with stock price. Although, the stock price should basically reflect the entity of the enterprise, the market price of stock is usually not equal to the entity value. In fact, the market price of stock tends to fluctuate in such a way that can not be explain by economics or finance theory. We, then, assume here that the market price of stock contains observation noise, and we try evaluate the option price by eliminating such noise and estimating the entity value of stock. In this paper, we assume that the observation noise of stock price follows white Gaussian process, and the entity stock price follows log-normal process. Using the result of entity stock estimation, we propose an algorithm of option evaluation. After testing the algorithm by some simulation, we calculated the estimated entity price of stock and implied volatility of call option by using the real market data.
In this paper, we consider state feedback control of discrete event systems, where we assume that the set of control patterns is closed under union. A control specification is assumed to be given in terms of a predicate on the set of states. First, we introduce the notion of Γ-controllability of predicates. We then show that Γ-controllability is a necessary and sufficient condition for the existence of a state feedback controller which achieves the control specification. However, the given predicate is not necessarily Γ-controllable. In such a case, its supremal Γ-controllable subpredicate plays an important role in synthesizing state feedback controllers. So we present a necessary and sufficient condition for the existence of a Γ-controllable subpredicate, and derive a closed form expression of the supremal Γ-Controllable subpreicate when the set of states satisfying the given predicate is finite.
When we work in virtual space, we need input device for measuring spatial motion of a human hand. In this research, we use force display for input device which generates reaction force from virtual object. We developed an operation system of virtual space. The system supports easy construction of virtual environment. Using this system, we developed a program that enables two users to work simultaneously in the same virtual environment. Through experiments of deformation of virtual objects, effectiveness of cooperative work in virtual space is tested.
Firsly, a new consistency index is proposed to the pair-wise comparison matrix which is extensively used in AHP for obtaining the relative importances of alternatives. By computer simulation of AHP processes, the properties of the proposed index are examinined in comparison with the conventional index which is based on the maximum eigen value of a pair-wise comparison matrix. The result shows the new index has high correlation with the conventional index and its consistency check criterion is given. Secondly, based on the proposed consistency index, a new simple method is introduced for modification of the pair-wise comparison matrix with low consistency. The computer simulation results of modification processes show that the propsed method can reduce the deviation from an original pair-wise comparison matrix more than the conventional method such as one using the sensitivity of the above consistency index. In this sense, the proposed method makes much account of the decision maker's first opinion and does not require its large modification.
The diagnosis method using qualitative reasoning is one of effective techniques for diagnoses of complicated systems such as air conditioning systems of buildings. In this technique, the target system is modeled qualitatively, and possible behaviours of the target model are generated by the qualitative reasoning. By comparing the results of reasoning with the real measured values, a part in failure is identified. However, the qualitative reasoning has its short-comings. That is, the possible behaviours of the model tend to increase enormously because of ambiguity of qualitative reasoning. To solve this problem, we propose the stochastic qualitative reasoning method in this paper. In this method, the existence probability is introduced. The states that have small probability are eliminated through the process of qualitative reasoning. Moreover the agreement degrees between results of reasoning and real measured values are calculated by existence probability of each states, so the cause of failure is identified in justifiable order. We have applied this method to fault diagnosis of a building air conditioning system practically and confirmed its effectiveness.
In excavating tunnels for subways and power cables, shield tunneling machines, which have many cutters on their cutter planes, are used. The author previously proposed, for the machines, an effective detection system which detected anomalous objects ahead, to prevent the cutters and the cutter plane from getting damaged and also to avoid the construction delay caused by the accident. The system made use of sonic transmitting and receiving transducers on the cutter plane to gather information on the anomalous objects. Not many observation data being available in the detection, optimal observation policy is practically of a great importance. The paper discusses strictly the optimal location of the receiving transducers on the cutter plane and also the optimal observation policy which takes into account the rectlinear and revolving motions of the cutter plane accurately. This consideration will be done for two cases. That is, the case where each acoustic sensor acts as one transmitting or receiving transducer and the case where each sensor has the two functions simultaneously.
New grasping pressure distribution sensor that has the structure that wire-electrodes are sewn in pressure conductive rubber sheet was developed. Some experiments to measure characteristics and to test efficiency of this sensor were also done.
In this paper, we define Lyapunov-like function and show, through an example, that the controller designed by using it stabilizes globally and asymptotically a control system with mismatched uncertainties.
In this paper, we present a new compliance control strategy for a robot manipulator to interact with its uncertain environment. Our strategy considers the manipulability explicitly in designing the robot's compliance. Computer simulations of the crank rotation task show that, our approach results in smaller interaction forces and better positional accuracy compared to the conventional compliance control.