Fuzzy sets operations should be considered when expert's subjectivity and a natural language are employed in system reliability analysis since natural language expressions of the analysis result are dependent on expert's subjectivity. This paper is concerned with considering fuzzy sets operations from the viewpoint of the fuzzy reliability analysis. In this paper parameterized t-norm and t-conorm are selected as fuzzy sets operations which can cover the range from the least t-norm(a drastic product) through the greatest one (a logical product) and the range from the least t-conorm (a logical sum) through the greatest one(a drastic sum), respectively. From the viewpoint of system reliability analysis the least t-norm and t-conorm mean the most optimistic operation, and the greatest t-norm and t-conorm mean the most pessimistic one. That is, the parameterized fuzzy sets operations can make the most optimistic model through the most pessimistic one for reliability analysis. Finally this paper compares analysis results among three kinds of subjective reliability analyses by an example and discusses the meaning of the parameterized operations from another point of view. The example also shows that parameterized operations reflect expert's subjectivity and intuition about reliability information based on his experience. Subjective reliability analysis is performed by introducing expert's subjectivity to fuzzy sets operations as well as to reliability estimate.

View full abstract

Recently, a lot of methods for transforming fuzzy multiple objective linear programming(MOLP) problems intocrisp MOLP problems have been reported. However, a transformed crisp MOLP problem results in an extremely lage number of constraints and decision variables. We feel that is the main disadvantage of these approaches. In this paper, we propose that a maximizing MOLP problem with trapezoidal positive fuzzy coefficients can be transformed into a crisp MOLP problem without an increase of only twice the number of constraints by using the two indices based upon the inequality relation between trapezoidal fuzzy numbers. Also, we demonstrate the usefulness of the proposed method.

View full abstract

In the ordinal structure model for fuzzy reasoning, all the fuzzy inference rules are described in one dimensional space for each input and output.Coordination of the rules are done with weights on the rules. The model makes it easy to apprehend the correspondence of the fuzzy inference rules of human beings. For deciding priorities of assembling products using the ordinal structure model, the weights on the inference rules for such numerous factors as deadlines, number of unavailable parts, working hours, etc. should be obtained. Since the experts in production management have some of the weights in mind unconciously, extraction of the weights has been a hard work. The weights for deciding the priorities of assembling products are considered in this paper as the knowledge to be acquired. This paper presents a new knowledge acquisition support system with a presentation of inferred modifications for acquiring the knowledge. This paper also presents the "PRIMITIVE CURVE" to infer proposals for correcting the pre-input weights. An experiment is performed with one of the authors assumed to be an expert to show the feasibility of the system. The experiment uses actual data of a production line.

View full abstract

This paper proposes a multi-class classification method using fuzzy if-then rules and examines its performance under various parameterizations in fuzzy inference. The fuzzy if-then rules employed in this paper are, for example, "If x_<p1> is small and x_<p2> is large then x_p belongs to Class 1 with CF=0.8" where CF is the grade of certainty of the rule. First we show a simple method for generating fuzzy if-then rules from training patterns. Our approach requires neither timeconsuming iterative computations nor complicated learning procedures. Next we show a fuzzy inference method for inferring the class of an unknown pattern. By computer simulations, we investigate the relation between the classification power of fuzzy if-then rules and various parameterizations in the fuzzy inference. For example, we examine the dependency of the classification power on the number of fuzzy subspaces, the shape of fuzzy sets and the type of operators. Last, by computer simulations on the iris data in Fisher, we compare the performance of the proposed method with that of the neural networks trained by the back-propagation algorithm.

View full abstract

The research reported here sought to validate the γ-model, a fuzzy-set technique for predicting attraction emotions. This computational model calculates the attractiveness of objects based on several variables, including appealingness, unappealingness, and familiarity. Ratings on all of the variables are combined using a fuzzy-set aggregation method that mimics human decision making. Eight subjects viewed each of 48 computer images of rooms, and rated them according to seven general adjectives and their degree of like and dislike. In addition, subjects rated the floor, walls, and ceiling of each room for its familiarity and appealingness/unappealingness. For subjects whose ratings of familiarity and appealingness/unappealingness were self-consistent, the γ-model predicted their performance about as well as standard regression techniques. Further investigation revealed that the γ-model is especially sensitive to changes in the dislike emotion. The results suggest that the γ-model can help us understand individual differences in attraction emotions.

View full abstract

A fire warning system is an indispensable system to protect human life, property and living conditions. On the other hand, false alarms cause us to miss real fires by switching off the system. For these social needs, we aim to develop a system which is to detect a fire at an earlier stage and then to give a highly reliable judgement. To realize such a system, this paper explains that a fuzzy expert system is effective due to the application of knowledge acquisition, and an inference mechanisms. That is, we propose to extract characteristic value by fuzzy clustering, to judge fire at an earlier stage by fuzzy reasoning, and to alarm gradually by graphics and voice synthesis. This prototype system has been installed in a real building and proves that it is effective in reducing false alarms from the sensed data. For fire alarm, we confirm the system's safety from experimental data. By using fuzzy theory, we can judge and act in an ambiguous situation before the final decision is made, and avoid unnecessary panic.

View full abstract

In this paper, decision making problems arising from optimal operation planning of an electric furnace for refining aluminium are formulated as multiobjective mixed integer programming problems by incorporating fuzzy theory together with multiple objectives rather than single objective. By considering the imprecise or fuzzy nature of human judgments, the fuzzy goals of the decision maker for each of the objective functions are also introduced. Then approximated solutions for the formulated problems are derived through the simulated annealing method for solving the general combinatorial optimization problems. In order to decrease the difficulties for the determination of appropriate parameter values in the simulated annealing method, an interactive parameter determination scheme as well as interactive decision making methods for solving the formulated problems are proposed. On the basis of the proposed methods, an interactive decision support system is developed on the workstation and the feasibility and efficiency of both the proposed methods and the corresponding decision support system are demonstrated via several numerical examples.

View full abstract

It is desirable to be able to express behavioral sequences which include the ambiguous state recognition when trying to represent "human-like" behavior-decision abilities in the intelligent robot. On this research, we try to express the macro behavior-decision algorithm close to the one which humans use every day by utilizing fuzzy algorithms capable of expressing sequence flow and handling both crisp and fuzzy processing. In this paper, we propose the modified fuzzy algorithm which has the algorithm tuning function based on fuzzy algorithms. The construction method of the behavior-decision algorithm for autonomous mobile robots using the modified fuzzy algorithm is also described in this paper. The main features of this method are as follows. 1) Allows the flexible and macro knowledge representation including behavior-decision sequences through the utilization of fuzzy algorithms. 2) Allows the definition of fuzzy concepts expressed by using fuzzy-based frames and fuzzy states expressed by the modified fuzzy algorithm. 3) Possesses the algorithm tuning capability based on the threshold value controlled by a behavioral purpose. Finally, we report some results of computer simulations and locomotive experiments concerning an evaluation of this method supposed simple in-door environments.

View full abstract

When an expert controls objects, first of all, he decides a rough input using global knowledge of himself concerning the control objects. Next, he estimates the control result using the rough input and modifies it a little in order to satisfy control objectives. His decision process about an optimal input to the control object can be divided into two different stages. This paper proposes a new fuzzy method, Two-Stage Fuzzy Control, which consists of two different inference stages. The first stage of Two-Stage Fuzzy Control corresponds to the expert's decision about the rough input and the second stage of it corresponds to the expert's modification of the rough input. Further, the second stage is also regarded as one of the different tuning method from conventional ones in fuzzy control. We apply Two-Stage Fuzzy Control to a parking control problem using model car and discuss the simulation results by comparing with the conventional states evaluation fuzzy control.

View full abstract

In order to improve control ability of robot manipulator path control, not only design of a servo system with high quality but also trajectory planning based on the given path are very important task. Moreover, online observation of the servo system performance is necessary because output of the servo system might deviate from the desired path because of initial error and disturbance etc. In this paper, online fuzzy trajectory planning method is proposed. As, in this method, new representation method named "P representation" for the given path is utilized, the trajectory planning method is easily applied for multi-degree of freedom robot manipulator. Here, P representation means that the given path is represented by phase angle of given path with respect to each axis of work space. Online fuzzy trajectory planning and tracking control are carried out simultaneously, so tracking ability of the robot manipulator is improved compared with a case utilizing of offline trajectory planning. Finally, the effectiveness of the proposed method of online fuzzy trajectory planning is shown by experimental studies.

View full abstract

An Optimum Fuzzy Controller which is applicable to the optimum operation of train's running is described. In designing the controller, not only the operating experience of skilled drivers can be fully considered, but also the Fuzzy Control Theory can be properly combined with the Optimum Theory. By developing an on-site controller realized by a microcomputer system, field tests had been finished in P.R. of China. As a result, it was testified that the fuel oil consumption was apparently decreased on the premise of safety and punctuality of train's running. In order to make characteristics of the Optimum Fuzzy Controller more perfect, it has been improved with the help of advanced technology and supporting tool of fuzzy control in Japan. The simulation result shows that the improved controller can ensure the train's running in an optimum operating way even though actual disturbances are great.

View full abstract