This paper presents the results of a project presently under way at Texas A&M which focuses on the use of fuzzy logic in integrated control of manufacturing systems. The specific problems investigated here include diagnosis of critical tool wear in machining of metals via a neuro-fuzzy algorithm, as well as compensation of friction in mechanical positioning systems via an adaptive fuzzy logic algorithm. The results indicate that fuzzy logic in conjunction with conventional algorithmic based approaches or neural nets can prove useful in dealing with the intricacies of control/monitoring of manufacturing systems and can potentially play an active role in multi-modal integrated control systems of the future.

We view fuzzy logic control technology as a high level language in which we can efficiently define and synthesize non-linear controllers for a given process. We contrast fuzzy proportional integral (PI) controllers with conventional PI and 2D sliding mode controllers. Then we compare the development of fuzzy logic controllers (FLC) with that of knowledge-based system (KBS) applications. We decompose the comparison into reasoning tasks (representation, inference, and control) and application tasks (acquisition, development, validation, compilation and deployment). After reviewing the reasoning tasks, we focus on the compilation of fuzzy rule bases into fast access lookup tables. These tables can be used by a simplified run-time engine to determine the FLC's crisp output for a given input. Finally we illustrate the application of FLC technology in a hierarchical architecture to control a complex power plant for heavy vehicles.

The paradigm of fuzzy modelling entails development of relationships (dependencies) between the linguistic entities defined for system's variables. The key feature of the fuzzy models pertains to their significant flexibility so they could easily be modified to comply with the principle of incompatibility. Considering the existing panoply of fuzzy models one can easily conclude that most of them are embraced under an umbrella of a single conceptual structure. From a functional point of view this structure is perceived as a combination of the two conceptual interfaces and a single processing block aimed at developing calculus of the linguistic labels. The interfaces produce all the links that are necessary to combine the physical (numerical) level of the real-world system with that of a conceptual character realized within the fuzzy model and articulated at the level of the linguistic entities. The presentation will address the main methodological aspects concerning these functional components with a particular emphasis placed on the associated design principles. The main issues dominating the design of the interfaces pertain to the implemented level of information granularity, optimality of linguistic labels, and linguistic-to-numerical transformations. The processing level of the fuzzy modelling will be considered through the use of fuzzy neural networks. These distributed computing structures are highly heterogeneous as they are constructed with the aid of several distinct types of logic-oriented neurons. The advantages of the fuzzy neural networks such as an implicit scheme of knowledge encapsulation that is carried out there will be discussed in detail.

Fuzzy expert systems can be developed for the effective use of management within the domains of concern associated with Operations Research and Management Science. These models are designed with: (1) expressive powers of representation embedded in linguistic variables and their linguistic values in natural language expressions, and (2) improved methods of interference based on fuzzy logic which is a generalization of multi-valued logic with fuzzy quantifiers. The results of these fuzzy expert system models are either (1) approximately good in comparison with their classical counterparts, or (2) much better than their counterparts. Moreover, for fuzzy expert systems models, it is only necessary to obtain ordinal scale data. Whereas for their classical counterparts, it is generally required that data be at least on ratio and absolute scale in order to guarantee the additivity and multiplicativity assumptions.

The development and analysis of a fuzzy logic rule based system capable of improving the quality of printed picture from video images is introduced in this paper. This will be achieved via fuzzy logic by controlling video image attributes such as color, sharpness, brightness and contrast. Each printed image is given a score (S) by human experts and then this result is stored in a database which consists of the corresponding video image attributes and video printer settings. From this database a set of fuzzy IF-THEN rules are derived to maximize the score S over a large set of printed pictures which results in minimizing the cost function.

This paper describes an application of fuzzy logic to an industrial temperature control system. It is a hybrid system of an advanced 2-degree-of- freedom PID controller unit and a fuzzy logic unit. The PID controller functions as a main control component while the fuzzy logical unit functions as a disturbance compensator. The 2-degree-of-freedom PID controller uses a feed forward operator to enhance the system's response to a desired value and allows simultaneous adjustments on target tracking and disturbance response. The fuzzy logic unit enables a simultaneous reduction of the disturbance recovery time and the overshoot amount. Both PID and fuzzy control parameters are tuned automatically, simplifying the user interaction. This hybrid control system has been successfully used in various industrial applications, and some of them are described in this paper.

During the last five years, Fuzzy Logic has gained enormous popularity, both in the academic and industrial worlds, breaking up the traditional resistance against changes thanks to its innovative approach to problems formalization. The success of this new methodology is pushing the creation of a brand new class of devices, called Fuzzy Machines, to overcome the limitations of traditional computing systems when acting as Fuzzy Systems and adequate Software Tools to efficiently develop new applications. This paper aims to present a complete development environment for the definition of fuzzy logic based applications. The environment is also coupled with a sophisticated software tool for semiautomatic synthesis and optimization of the rules with stability verifications. Later it is presented the architecture of WARP, a dedicate VLSI programmable chip allowing to compute in real time a fuzzy control process. The article is completed with two application examples, which have been carried out exploiting the aforementioned tools and devices.

From its conception by Professor Lotfi A. Zadeh in the early '60s, Fuzzy Logic has slowly won acceptance, first in the academic world, then in industry. Its success is mainly due to the different perspective with which problems are tackled. Thanks to Fuzzy Logic we have moved from a numerical/analytical description to a quantitative/qualitative one. It is important to stress that this different perspective not only allows us to solve analysis/control problems at lower costs but can also allow otherwise insoluble problems to be solved at acceptable costs. Of course, it must be stressed that Fuzzy Systems cannot match the computational precision of traditional techniques but seek, instead, to find acceptable solutions in shorter times. Recognizing the enormous importance of fuzzy logic in the markets of the future, SGS-THOMSON intends to produce devices belonging to a new class of machines: Fuzzy Computational Machines. For this purpose a major research project has been established considering the architectural aspects and system implications of fuzzy logic, the development of dedicated VLSI components and supporting software.

This paper presents the classification of fuzzy dynamic systems and fuzzy linguistic controllers (FLC) into standard types (TYPE 1 through TYPE 7). The need, utility value, and the logic behind this classification are given. The proposed classification is the result of studying many known examples of FLC applications. The impact of this classification to new designs and to the improved performance of classical and modern control systems is an important consideration.

The variety of illumination conditions in outdoor environments--in space or on earth--does not allow to use simple video cameras as a single perception means. A system composed of complementary sensors and sensitive to multiple frequencies is necessary when the system has to be efficient and reliable for any illumination and visibility conditions. On the other hand, geometrical information (edges, surfaces, etc.), which is extracted from usual perception systems, is not always sufficient to assure the safety of a navigation task. The physical properties of the surfaces such as the roughness and the dielectric constant (also related to the water content) represent interesting information since they allow to evaluate the navigability of the observed terrain: smooth surface, rugged surface, stones or soil, metallic surfaces, vegetation. etc. Unlike geometrical properties, these physical parameters are independent of the observation point and illumination conditions. Also, when associated to geometrical parameters, they should facilitate the location of a mobile vehicle by helping object recognition tasks. We defined a perception system based on the exploitation of theoretical perception models using fuzzy logic rules. This system should assure the perception of the scene for any illumination conditions and extract geometrical and physical surface properties. This new system should therefore enhance the navigation autonomy and reliability. We present the advantages of the developed approach and show the first testing results.

Electric Power Research Institute has conducted a survey of recent advances and trends in Japanese thermal power plants during summer 1991. This paper summarizes the survey, and introduces success stories and lessons learned in Japan. The survey has found many applications and wide-spread implementations of fuzzy logic technology such as: fuzzy-logic schedulers for plant transient operations, fuzzy-expert tuners of dynamic control systems, and fuzzy- algorithmic operation guidance systems for major plant equipment. It has been also found that applications of fuzzy logic technology caused drastic changes in functions of human operators in the loop [man-machine interface], database management and preventive plant maintenance, and control room design.

Many authors are currently investigating fuzzification of self-organizing algorithms. This paper discusses some problems that can be expected during attempts to generalize Kohonen's self-organizing feature map (SOFM) as it is used for feature extraction and visual display. We review three methods for solving each of these two problems: principal components analysis; Sammon's algorithm; and Kohonen's SOFM algorithm. Then we present a number of numerical examples that illustrate some difficulties with the SOFM approach. We propose a modification of SOFM that extracts feature vectors in q-space from data in p-space. However, since the coordinates of the extracted points are constrained via logical connectivity to a display lattice in q-space, the resultant features are not particularly good lower dimensional representations of the data they attempt to mimic. Our metric topological preservation index suggests that Extended SOFM does not preserve topological relationships nearly as well as principal components or Sammon's algorithm.

We introduce the concept of higher order criteria in the decision making problem. A prototypical manifestation of these types of criteria occurs in situations in which we desire to satisfy a criteria if it is possible without sacrificing the satisfaction to other primary criteria. We show how these types of criteria can be used to help model the process of satisficing.

A few existing applications of fuzzy logic in medicine are briefly described and some potential applications are reviewed. The problem of classification of patient states and medical decision making is discussed more in detail and illustrated by the example of a fuzzy rule based model developed to elicit, analyze and reproduce the opinions of multiple medical experts in the case of arterial hypertension. The goal was to reproduce the average coded answers using an adequate fuzzy procedure, here a fuzzy rule. State categories and the initial set of experimental parameters were defined according to medical practice. The fuzzy set membership functions were then assessed for each parameter in each category and a small subset of representative and pertinent parameters selected for each question. The data were split into two sets of 50 patient files each, the calibration set and the validation set. Two evaluation criteria were used: the sum of squared deviations and the sum of deviations. Fuzzy rules were then sought that reproduced the target, which was the average coded answer. Only one fuzzy rule `and' appeared to be necessary to describe the patient state in a continuous way and to approach the target as closely as the majority of experts.

This paper presents a discussion of the potential of Fuzzy Logic Technology (FLT) for the IVHS (Intelligent Vehicle/Highway System) program. After a review of the IVHS program, some relevant roles of FLT are highlighted and illustrated with examples from recent works on the subject.

Increasing complexity of systems requires improved support capabilities. Automation controls the support costs while meeting the growing demands at the same time. Proteus is a firm-wide proactive problem management system with automated advisory capabilities. Proteus non-obtrusively accumulates troubleshooting expertise and quickly recycles it by combining case-based reasoning with text retrieval and fuzzy logic pattern matching. It has linear on-line and sub-quadratic preprocessing computational time complexities.

Consumer preference models are widely used in new product design, marketing management, pricing and market segmentation. The purpose of this article is to develop and test a fuzzy set preference model which can represent linguistic variables in individual-level models implemented in parallel with existing conjoint models. The potential improvements in market share prediction and predictive validity can substantially improve management decisions about what to make (product design), for whom to make it (market segmentation) and how much to make (market share prediction).

If tolerances are defined by crisp numerical values, several vehicles based on probabilities and classical statistics exist to conduct a process capability analysis. If data are categorical, however, and if they are obtained from subjective evaluations, the existing methods are inappropriate. Furthermore, if specifications are set in lexical terms or are loosely defined, current approaches are impossible to implement. This paper applies fuzzy logic theory to study process capability in the presence of uncertainty and categorical data. Examples are discussed using TIG welding experimental data.

One common assumption of models to evaluate the cost of variation is that the quality characteristic can be approximated by a standard normal distribution. Such an assumption is invalid for three important cases: (a) when the random variable is always positive, (b) when manual intervention distorts random variation, and (c) when the variable of interest is evaluated by linguistic terms. This paper applies the Weibull distribution to address nonnormal situations and fuzzy logic theory to study the case of quality evaluated via lexical terms. The approach concentrates on the cost incurred by inspection to formulate a probabilistic-possibilistic model that determines cost savings due to variance reduction. The model is tested with actual data from a manual TIG welding process.

The Department of Energy is making significant improvements to its nuclear facilities as a result of more stringent regulation, internal audits, and recommendations from external review groups. A large backlog of upgrades has resulted. Currently, a prioritization method is being utilized which relies on a matrix of potential consequence and probability of occurrence. The attributes of the potential consequences considered include likelihood, exposure, public health and safety, environmental impact, site personnel safety, public relations, legal liability, and business loss. This paper describes an improved method which utilizes fuzzy multiple attribute decision methods to rank proposed improvement projects.

The paper deals with a new approach to the learning of fuzzy rules. It suggests a solution to one of the problems of crucial importance for the learning of fuzzy rules by back propagation- -the issue of estimation of the initial values of the unknown parameters. We introduce the method of clustering via the mountain function to identify the most important rules. Those are the rules that are associated with higher values of the peaks of the mountain function. From the centers of the clusters that are obtained by the mountain function method are determined the initial estimates of the parameters of the reference antecedent and consequent fuzzy sets of the rules. In the next step the method of back propagation is used for more precise identification of those parameters.

Proper evaluation of evidence from multiple sources is often difficult, especially when some or all of that evidence is ill-defined. A combination of traditional probabilistic methods and fuzzy set theory provides a way to combine such evidence in a statistically meaningful manner. This paper shows that appropriate versions of Bayes' Theorem apply to fuzzy events.

A neural-fuzzy system combined supervised and unsupervised learning to find and tune the fuzzy-rules. An additive fuzzy system approximates a function by covering its graph with fuzzy rules. A fuzzy rule patch can take the form of an ellipsoid in the input-output space. Unsupervised competitive learning found the statistics of data clusters. The covariance matrix of each synaptic quantization vector defined on ellipsoid centered at the centroid of the data cluster. Tightly clustered data gave smaller ellipsoids or more certain rules. Sparse data gave larger ellipsoids or less certain rules. Supervised learning tuned the ellipsoids to improve the approximation. The supervised neural system used gradient descent to find the ellipsoidal fuzzy patches. It locally minimized the mean-squared error of the fuzzy approximation. Hybrid ellipsoidal learning estimated the control surface for a smart car controller.

In this paper, fuzzy logic is used to derive knowledge combination rules for heuristic search algorithms. These algorithms are especially designed for expert systems addressing complex problems whose solution space is too large to be fully explored. Search algorithms using heuristic knowledge can reduce the exploration of the solution space while preserving the quality of the proposed solutions.

The extended Hough transform permits weight functions of arbitrary type and complexity to help guide the choice of a `regression' line in polar coordinate space. This paper suggests that this transform may be helpful in locating the best linear approximation to gaps and areas of conflict in fuzzy rule bases. Using the sliding mode approximation of a fuzzy controller as an example, this paper shows how global properties of the rule base can be used to help guide the search for good approximations. The notion of `representativeness' of centroids and its effect on regression via the Hough transform is also considered. Finally, a different approach based on OWA operators is discussed briefly.

This paper presents an expert knowledge-based decision support system for capital investment risk analysis. The theoretical foundation of the DSS is Possibilistic Evidence which connects the Possibility Theory and the Theory of Evidence. This paper illustrates the fundamental principles of the expert knowledge representation, the framework of inference with possibilistic evidence and the architecture of the system. A typical case of capital investment risk analysis is introduced to demonstrate the application of the system.

The Fuzzy Logic Expert System Scheduler (FLES) is a unique, on-line, interactive shop floor scheduler that is designed to produce detailed, realistic schedules for day-to-day production management. The user can exercise the control of FLES to produce scheduling assignments over short or long term scheduling horizons or to simulate different plant capacity conditions to analyze their effect on future work plans. The unique and proprietary feature of FLES is its `Decision Engine', a fuzzy knowledge base system that models the reasoning process of a human expert is used to give job releasing and job dispatching decisions. Expert knowledge in terms of fuzzy production rules represented by the use of linguistic variables. The values of these linguistic variables are defined by context dependent fuzzy sets whose meanings are specified by graded membership functions.

Each year, the Army National Guard (ARNG) reviews its entire force structure and, based upon changing national requirements, determines which units, in which states, will be inactivated, activated, or reallocated. These decisions are based upon a fairly large set of criteria and include: overall manpower goals, socioeconomic factors, and unit performance characteristics. In order to facilitate these decisions, a fuzzy expert system is currently being built for the ARNG using a fuzzy logic inference engine based upon the fuzzy decision-making methodologies of Bellman & Zadeh as extended by Yager, and the analytical hierarchical method of Saaty, which normally uses paired comparisons. Incorporated within the force realignment decision support system, is a method of capturing decision maker preferences, which does not require paired comparisons and is (1) rapid, and (2) forces transitivity in the resulting fuzzy relative importance measures.

Fuzzy logic technology has been applied to control problems with great success. Because of this, many observers fell that fuzzy logic is applicable only in the control arena. However, business management problems almost never deal with crisp values. Fuzzy systems technology--a combination of fuzzy logic, fuzzy mathematics and a graphical user interface--is a natural fit for developing software to assist in typical business activities such as planning, modeling and estimating. This presentation discusses how fuzzy logic systems can be extended through the application of fuzzy mathematics and the use of a graphical user interface to make the information contained in fuzzy numbers accessible to business managers. As demonstrated through examples from actual deployed systems, this fuzzy systems technology has been employed successfully to provide solutions to the complex real-world problems found in the business environment.

Research in the last decade emphasized the potential of designing adaptive pattern recognition classifiers based on algorithms using multi-layered artificial neural nets. The greatest potential in such endeavors was anticipated to be not only in the adaptivity but also in the high-speed processing through massively parallel VLSI implementation and optical computing. Computational advantages of such algorithms have been demonstrated in a number of papers. Neural networks particularly the self-organizing types have been found quite suitable crisp pattern for clustering of unlabeled datasets. The generalization of Kohonen-type learning vector quantization (LVQ) clustering algorithm to fuzzy LVQ clustering algorithm and its equivalence to fuzzy c-means has been clearly demonstrated recently. On the other hand, Carpenter/Grossberg's ART-type self organizing neural networks have been modified to perform fuzzy clustering by a number of researches in the past few years. The performance of such neuro-fuzzy models in clustering unlabeled data patterns is addressed in this paper. A recent development of a new similarity measure and a new learning rule for updating the centroid of the winning cluster in a fuzzy ART-type neural network is also described. The capability of the above neuro-fuzzy model in better partitioning of datasets into clusters of any shape is demonstrated.

Neural networks known for almost a half-century deal with the complexities of the natural world data by generalizing from prototypes or samples in much the same as humans extrapolate from one experience to similar situations. Neural networks have been studied for fault diagnosis and health monitoring of space operations and have been extended to include time correlated data and create the Space-Time Neural Network. Neural networks have been combined with fuzzy logic to investigate the adaptive decision making systems for translational and rotational control. In this paper, we describe some of the space applications of fuzzy logic and neural networks including tether skip-rope identification using Space-Time Neural Network. We also describe the tools developed at Johnson Space Center to support such investigations.

This paper overviews four combinations of fuzzy logic, neural networks and genetic algorithms: (1) neural networks to auto-design fuzzy systems, (2) employing fuzzy rule structure to construct structured neural networks, (3) genetic algorithms to auto-design fuzzy systems, and (4) a fuzzy knowledge-based system to control genetic parameter dynamically.

The paper introduces a neural network-based model of logical connectives. The network consists of two types of generic OR and AND neurons structured into a three layer topology. The specificity of the logical connectives is captured by the network within its supervised learning. Further analysis of the connections of the network obtained in this way provides a better insight into the nature of the connectives for fuzzy sets; in particular the analysis can look at their non-monotomic and compensative properties. Numerical studies including the Zimmermann-Zysno data set illustrate the performance of the network.

A new approach to fuzzy distance and restriction measures is used to obtain the appropriate orientations of the links for avoiding obstacles in the robot trajectories. This approach eliminates the classical task of solving highly coupled, nonlinear equations describing the ill- posed inverse problems of multilink robot motion at a much less demanding computational time. Such clear advantage of fuzzy logic based adaptive controller are illustrated by simulation results of guidance of a multilink robot in target positioning and trajectories tracking. The simulation results involve a three-link robot arm with capability of moving from one position to any desired position and tracking a defined trajectories accurately. A modified fuzzy rule based distance measure allows the robot to follow trajectories within hitting the obstacles in the path. The simulation results indicate the advantage of fuzzy logic based adaptive controllers in multiple criteria decision-making tasks.

Over the last decade or so, significant advances have been made in two distinct areas: fuzzy logic and computational neural networks. The theory of fuzzy logic provides mathematical strength to compare the uncertainties associated with human cognitive processes, such as thinking and reasoning. Also, it provides a mathematical morphology to emulate certain perceptual and linguistic attributes associated with human cognition. On the other hand, the computational neural network paradigm has evolved in the process of understanding the incredible learning and adaptability of biological neural mechanisms. Neural networks replicate, on a small scale, some of the computational operations observed in biological learning and adaptation. The integration of these two fields, fuzzy logic and neural networks, has given birth to an emerging paradigm--the fuzzy neural networks. The fuzzy neural networks have the potential to capture the benefits of the two fascinating fields, fuzzy logic and neural networks, into a single capsule. The intent of this paper is to provide an introductory look at this emerging research field of fuzzy neural networks.

A method of tuning a fuzzy logic controller (FLC) by a genetic algorithm (GA) is proposed for lane following maneuvers in an automated highway system. The GA simultaneously determines the shape of membership functions, number of rules, and consequent parameters of the FLC. The GA approach operates on binary representations of FLCs and uses an expression for a fitness score to be maximized, which takes into account the tracking error, yaw rate error, lateral acceleration error, rate of lateral acceleration, front wheel steering angle, and rate of front wheel steering angle, to find an optimal controller. Apriori knowledge about both the physical application and FLCs is incorporated into the design method to increase the performance of the design method and the resulting controller. The controllers designed by this method are compared in simulation to a conventional PID controller, a frequency shaped linear quadratic controller, and previously designed FLCs tuned manually.

We have added a real-time interactive fuzzy reasoning system and neural network simulator to the MAX real-time music programming language. This environment allows us to quickly prototype and experiment with Neural, Fuzzy, and Neuro-Fuzzy systems for control of real- time musical processes. In this paper we introduce our tools and discuss musical contexts that call for the adaptive and generalization capabilities of these systems.

The suspension system of a passenger car provides isolation between the occupants in the car and the road surface. The three goals of the suspension system are to provide ride isolation from vibration, limit suspension travel, and maintain road holding characteristics. Each of these three goals conflicts with the others. Thus, the controller must be designed to attain each goal to some extent. This paper proposes the use of a linear quadratic regulator and a fuzzy controller to maintain the ride isolation of a loosely sprung, lightly damped passive suspension while improving the handling characteristics of the vehicle. The suspension performance as pertains to ride isolation can be studied using a simple quarter car model of a suspension system. However, the handling characteristics and the coupling between each quarter of the suspension system must be studied using a full car model. Thus, this paper uses both a quarter car and a full car model to study the performance of suspension systems. The performance of the suspension systems is evaluated by running simulations of the systems subjected to both discrete and random road inputs. This paper shows that an active suspension using a linear full state feedback controller performs better than a passively suspended vehicle. The optimally controlled active suspension system is also compared to a fuzzy controlled active suspension system and the results are discussed.

An innovative design of a dynamic neural network architecture that is able to first learn and then utilize fuzzy-like `IF-THEN' rules is presented in this paper. Each fuzzy neuron in the network represents a compositional rule of inference that defines the relationship between a particular premise and the corresponding consequence. The neural network first determines the similarity between a neural input (a discretely sampled fuzzy set) and the feedforward synaptic weights (accumulated knowledge-base). The `best' definition of the input is selected by competition arising from the dense feedback between the neurons. A satisfactory conclusion is reconstructed in weighted feedforward outputs from the `winning' neuron. The knowledge-base is updated by an unsupervised learning algorithm that adapts the feedforward weights assigned to both the neural inputs and outputs. An example of how this dynamic neural network can be used to perform fuzzy-like inference rules for the navigation of an autonomous vehicle through an unstructured environment is used to illustrate these notions.

Liquid metal spray forming is a relatively new process for near-net-shape manufacturing. This process involves melting of the metal, ejection of the molten metal from the crucible in which it was melted, and atomization of the molten metal in a high-pressure gas atomizer. The resulting droplets are collected on a moving substrate where consolidation and solidification occur. This paper will discuss the methods used to combine sensor feedback and process knowledge into a fuzzy logic inference engine which can augment, if not replace, the manual functions currently performed by the operator. The paper will also describe the development of a custom optical sensor which simulates the operator's visual feedback. It monitors the part while it is being formed, and simultaneously determines the part shape, the rate of growth of the part, and the surface roughness of the part.

This paper will discuss fuzzy control and PID control to the application of rice wine production.

In this paper we present an architecture of fuzzy neural networks for adaptive aggregation in decision making. The excitatory and inhibitory mechanism of computational neurons is employed to model the competitive and cooperative behaviors among alternatives in the process of adaptive aggregation. An application of this architecture into the field of abductive diagnosis with causal knowledge is discussed in detail.

In many fuzzy system applications, the most difficult and time consuming problem is to built the fuzzy rule base. Usually, to build fuzzy rule base depends on a domain expert to reflect his experience. But for a complicated system, it is sometimes difficult for an expert to describe clearly the causal relationships among those linguistic variables. To overcome such a problem, a dense connectionist structure of artificial neural network, called as NN-Fuzzy Inferencer (NNFI), is constructed to implement the fuzzy inference. This NNFI incorporates the effects of neural network and fuzzy inference. It is trainable and gets a more desired output value than backpropagation neural network does. The idea of the NNFI architecture is driven from the traditional fuzzy inference method. It can avoid not only the difficulty that for a designer to define the casual relations between the input variables and output variables, but also determine the membership function for each linguistic value. Furthermore, the system will generate the weighting coefficients in antecedent part and consequent part respectively in every fuzzy rule.

In this paper, a linguistic approach to the problem of object recognition is outlined and fuzzy logic solution is given for this formulation. The problem under investigation is divided into two sub-problems: (1) feature extraction from a digitized image, and (2) matching features extracted against a set of pre-defined objects. Using linguistic approach, solving the problem of feature extraction yields a set of linguistic descriptions of an object from a particular digitized image. These descriptions will be matched with known patterns to derive a decision on the shape of the object. The use of fuzzy logic in matching patterns has been shown to provide satisfactory results in Monte Carlo simulations, especially in the case that there is ambiguity in the pictures. This ambiguity is often caused by a lack of extracted features in matching a pattern resulted from some viewpoint at the object or from a picture of poor quality.

A Fuzzy Deductive Database System has been developed that diagnoses the disease of a patient of pediatric age group with convulsion. The entire system has been encoded in Prolog to provide a rich query interface to it. In order to provide a uniform query interface to the deductive database, a fuzzy SQL-like (a subset of SQL) query interface to the system was developed. The scheme adopted provides a framework for combining different uncertainty management mechanisms to suit any particular application. This query interface to the database can be used to identify correlations among different symptoms and/or clinical or investigational findings among groups of patients which may suggest possible new rules. The ideas of fuzzy deductive databases have also been applied to the domain of high level VLSI synthesis. The fuzzy SQL interface developed, can be used to query the database of uncertain schedules in the dataflow graph, which can be used to arrive at an estimate of the number of functional units required to carry out all the operations. This information can be used by the main software performing scheduling, allocation and binding.