In orfer to construct the generalized tree structure for the classification samples, the method for learning the structure of the fuzzy decision tree based on Genetic Algorithm(FDTGA) is proposed in this paper. The proposed method has the following features; (1)Genetic Algorithm is applied to the tree structure learning, (2)The structure of the tree is indirectly encoded into the chromosomes, (3)It acquires multiple solutions during the learning. Thus, FDTGA is expected to construct compact and appropriate structures from the complex classification samples defined by the large nember of the attributes.In this paper, after the explanation of the proposing mehtod, we will attempt to acquire the structure of the tree and the fuzzy classification rules for the objective samples from the numerical simulations. Then, the results of the simulations will be compared with the other methods. In the results, we will consider the relation between the acquired rules and the samples, and the effectiveness of the proposed method will be shown.

View full abstract

This paper investigates a novel technique for constructing and evaluating neuro-models of pattern generation. The modeling approach consists of two steps : first a neural network is trained to learn a core concept expressed on fuzzy logic, and then the trained network is augmented with additional processing nodes and connections.The augmented netwoek is then tested on its ability to solve problems related to the core concept for which its was trained. We present results from applying our model to monotone-function generation, decision making and image generation.

View full abstract

In this paper, time-domain stability analyses on nonlinear control systems with various types of fuzzy controllers are discussed, in which a BIBO stability criterion and a global asymptotic stability criterion are given based on both boundedness and asymptoticity for solutions of the set of nonlinear inequalities with respect to characteristic quantities such as the quadrant components of three-dimensional input-output space of a fuzzy controller, positive or negative areas of an impulse response of a linear control object, and ultimate positive or negative bounds for input signals.Stability analyses for nonlinear control systems with a two-input or three-input and one-output fuzzy controller through four typical reasoning methods, i.e., the min-max-center of gravity method, the product-sum-center of gravity method, the functional reasoning method, and the Tsukamoto's method, are shown comparing the results due to the proposed method with those owing to other known stability criteria.As a result of the above, it is assured that the proposed method is systematically applied to every type of controllers with various reasoning methods and is also superior in stability ranges owing to the essential property of this proposed method.

View full abstract

This paper proposes a method to select essential elements in a huma nevaluation model using fuzzy measures and integrals, which considers relation among fuzzy measures of sets with more than two elements. Assuming that a combined element consists of composed elements, the proposed method selects not a combined element of which fuzzy measure is super-additive but composed elements which compose a combined element and the fuzzy measure of the combined element is super-additive. To this end, Increment Degree from fuzzy measures of composed elements to the fuzzy measure of a combined element, Average of Increment Degree, Necessity Degree considering both the Average Increment Degree of a combined element and its fuzzy measure which implies the importance of the element, are defined. The proposed method selects essential elements by the use of the relative Necessity Degree.The proposed method is applied to the evaluation of restaurants and the evaluation of the strength of wood and its usefulness is verified by comparing the proposed method with the conventional method. The results show that the proposed method select essential elements well in the evaluation when the number of elements is restricted.

View full abstract

At present, vibration control in architectural engineering is very much in the limelight. The reason why this is so is because the durability of a structure is improved by performing appropriate vibration control. Vibration control can give the structure the capability to bear complicated external forces, for example seismic forces, wind forces and so on.The authors are interested in a structure that is able to transform its own characteristics. When the spring constant of the structure is changed, the resonance point of the structure is changed. It is said that shifts in the resonance point can affect displacement caused by seismic forces.Therefore, we produced a small experimental equipment that is able to transform the characteristics of the structural model. The quipment has 2 pieces of concentration mass, variable springs and dampers. Those springs, dampers are virtual devices. Those are driven by DC servo motors. First of all, a mathematical model of the experimental equipment was obtained and vibration control methods were studied on the basis of the result. We used fuzzy control that was described by words. The control method avoids a resonance point. From the above, it is understood that fuzzy control is an effective method in vibration control.

View full abstract

Using fuzzy clustering and neuro-fuzzy techniques, was develop a new learning approach for tuning fuzzy rules under fuzzy singleton-type reasoning method. In this approach, using training data, we roughly design initial fuzzy rules based on fuzzy clustering algorithms before the learning of a fuzzy model, so that the fuzzy rules generated by a neuro-fuzzy learning algorithm are more reasonable and suitable for identifying the system model than the conventional method. Moreover, we show the efficiency of the developed approach by means of numerical examples.

View full abstract

A semantic analysis is recommendable for planning of the road networks. In this study, the image survey with colour description was taken to the road networks in Gifu city. The road is generally recognized as a particular route by the drivers as well as the road planners. Therefore, the road network may consist of the routes as a group. The colour image involved many semantic aspects. Fuzzy clustering is used to analyze the relationship between the image of the roads and the cultural climate of the city. The estimation of the image is estabilished by a neural network model. After consideration the connection of physical phenomena with the generated image, the method to know the mental climate of the city created by the road networks might be developed. The optimal combinations of colour images were founded by genetic algorithm

View full abstract

The traffic assignment is regarded as the most important estimation technique of traffic flow in urban network planning. In this study, traffic assignment models with fuzzy travel time are introduced. The applicability of the proposed model is investigated with empirical data. Firstly, the shape of the membership function for fuzzy travel time is considered from the data in the road traffic census and the road user questionnaire survey. Secondly, two different methods such as a removal model and a gravity model are proposed to represent fuzzy numbers. Both models estimate traffic flow with representative value of fuzzy travel time. They can be formulated as extensions of an equivalent mathematical programming for conventional atraffic assignment. Lastly, the models are applied to analyze the traffic flow on the network of Kyoto city as a real scale ezample. Particularly, the models are confirmed to be practical with consideration of the relation between observed values and estimated values for traffic in different types of roads.As the results of the study, it is concluded that the proposed models enable to describe the perception of drivers for travel time information and they can provide the practical estimation of traffic flows.

View full abstract

Recently, in the field of the outdoor structure design, it is increasingly be demanded alternatives have to be eaxmined in three dimensions even in initial stage of the development. A system which output three dimension prototype from designer's sketches has been constructed here based on such backgrounds. This paper outlines the usage and the specificaiton of this system introduction of an application case with the system is also presented. It has been found that the system completes it's task appropriately after simple appraisal is carried out for the figure which the system outputs. Considering the nonlinearity concerning design and the vague nature of knowledge, fuzzy inference was applied to calculate the value of the parameters which was used to construct the figure.

View full abstract

We have investigated some mathematical properties of De Morgan algebras and Kleene algebras which have some fixed points of unary operations up to now. In this paper, we introduce a new concept called fixed core which extends a fixed point, and give the characterization of De Morgan algebras and Kleene algebras by the fixed cores.

View full abstract

In this paper, we show how non-fuzzy scheduling problems can be reformulated in the framework of fuzzy due-data.The membership function of a fuzzy due-data assigned to each job shows the human user's satisfaction grade for the completion time of that job. Thus the fuzzy due-data is directly related to the eariliness and tardiness penalty in conventional scheduling problems. The fuzzy due-data can be also used for representing other scheduling criteria such as the makespan and the total flowtime. In this paper, we first desribe the concept of fuzzy due-date. Next we show that various scheduling criteria can be represented in terms of the membership function of the fuzzy due-data assigend to each job. That is, we show that many non-fuzzy scheduling problems can be reformulated through the concept of fuzzy due-date. Then we show how we can combine various scheduling criteria into a single measure using various shapes of membership functions for fuzzy due-date. The similarity and the difference between conventional scheduling problems and fuzzy scheduling problems are illustrated by computer simulations on several numerical examples.

View full abstract