We consider methods of constructing intelligent systems of geometric modeling and propose a technique and software for implementing these methods. The results obtained can be used in computer-aided design.The construction of high-efficiency systems for computer-aided design (CAD) is a problem of current interest. The efficiency of such systems can be greatly elevated by using the modern achievements of computational and constructive geometry [9], methods of intelligent automated interactive programming [3], intelligent knowledge base systems, expert systems, decision-making systems, and methods of multi-criterial optimization.The general situation in the area under investigation can be characterized by the level attained in these areas of theoretical and applied science. The most important particular problems of computational and constructive geometry are: the algorithmic aspects of geometry, data structures for representing geometric objects, optimization of geometric data structures, synthesis of solid geometric objects, visualization of geometric objects, the finite element method, optimal triangulation algorithms, representation of geometric knowledge, elimination of contradictions, and automation of geometric proofs and constructions.Among the Russian publications in the area of constructive geometry one may note [1, 2], which are devoted to the axiomatization of the problem of synthesizing 3-dimensional objects. On the basis of these investigations, in particular, an interactive graphical system has been constructed for synthesizing an intramachine model of a 3-dimensional object via instructions for the positioning of the base elements (primitives) in space and prescribing geometric operations. However, the correctness of the construction process can be evaluated only on the basis of visual criteria of the user, i.e., from the mapping of geometric images on the screen with no way of using the knowledge base and system of logical inference, and no way of avoiding contradictions or automating the system of geometric proofs and constructions. What has just been said holds also in regard to the most widely used foreign software for computer-aided design, such as AutoCAD, CADPlan, EUCLID [7], etc.One of the central problems of constructive geometry is the optimization of the inramachine representations of complicated geometric objects (binary trees representing expressions over primitives in an algebra with signature f~ = {N*, t.J*, \* }, and also optimization of ring list structures representing the surfaces of complicated geometric objects) and algorithms over these representations. In particular the paper [9] already mentioned is devoted to these problems; in this paper quite efficient algorithms are proposed for minimizing binary trees. Also devoted to this topic is [8], in which a rapid algorithm is proposed for carrying out refined set-theoretic operations over constructively defined geometric objects. However questions as to the correctness of data structures (inconsistency and completeness of descriptio...

We study current questions of the theory of design of interactive automated systems of technological manufacturing. We propose algorithms for organizing the "intelligent" interface of the user and the system. We give an outline far incorporating the intelligent model i,,tu the system environment.The efficiency of systems of automated design can be greatly increased by using methods of logical and functional programming, knowledge bases, expert systems, decision-making systems, various models of representation, completion and processing of geometric knowledge, and an intelligent interaction of the user with a computer.We shall study the problems that arise in developing design systems:A. The representation of knowledge and the synthesis of conceptual models;B. The development of an intelligent interface of the user with the system.Traditionally problems A and B have been studied independently, leading to the neglect of certain aspects of the problem. We shall consider a unified approach to the design of the user/system interface, whose essence reduces to determining a dialog scheme on the basis of a conceptual model for representing the knowledge of the subject area. We shall call this approach semantic (or conceptual) modeling and a system of design based on it intelligent. It is known that interactive software strongly affects all the components of a system, including the functional parts Thus it becomes important to develop a program environment that implements intelligent interface functions on the level of knowledge.We consider the problem of building a universal programming method--an Intelligent Dialog Editor (IDE)--designed for constructing dialog scenarios of user-defined type. The scenario (algorithm) and the dialog type depend on the knowledge available in the system on the subject area, as well as the description of the condition of the problem posed by the user. The use of the subject area model makes possible an automatic verification of the well-posedness of the problem and the construction of effective interactive algorithms for solving it.The degree of intelligence of dialog interfaces depends on the complexity (efficiency) of the logical deduction procedures used under conditions of incomplete input information. We shall require the IDE to have the ability to form lists of missing input data on the basis of a conceptual model and determine the values of the missing data by using logical and arithmetic procedures or by asking for them in the dialog independently of the reasons leading to the deficiency (inconsistency, uncertainty, overdeterminedness).We shall examine below the basic IDE algorithm. To verify the incompleteness of the list of input parameters of the problem we posit a method proposed by the authors for verbal representation and analysis of functional models of knowledge. We use expert data to verify the correctness of the values of the parameters in the IDE and a logical AND/OR dialog interaction to saturate the list of values of the input parameters of the IDE. The scenario of the ...

No abstract

Automation of intelligent programming is considered. A technology is proposed for the construction of intelligent program units based on verbal representation and analysis of functional models.One of the main goals in the theory of automatic programming is to develop technologies capable of constructing intelligent interactive systems, synthesizing algorithms, and generating solution methods for applied problems. The essence and the relevance of this topic are discussed in [1]. Algorithm synthesis and program generation rely on semantic models of the problem. The corresponding processes are called computation planning on semantic models [5].The bulk of computations in automatic programming systems are computations using functional models [3, 4]. In this paper, we propose methods of verbal description of functional models and examine the principles and methods of construction of intelligent program units (IPU) that resolve functional models. The IPU input is a list w (incomplete or redundant) of variable names (descriptors) of the functional model M with a distinguished goal parameter (descriptor) t.Resolution of the model M with a set of specifiers F includes proof of existence and construction of a derivation of the descriptor t from the elements of w. The derivation is a sequence P of descriptor resolutions.

No abstract