Development of Mathematical Models for Identification of the Category of the Technical Condition of Building Constructions on the Basis of Fuzzy Logic

Abstract

The paper contains justification for the relevance and necessity of developing expert systems for assessing the technical condition of construction projects. The description of the structure, construction methods and principles of operation of the fuzzy expert system for determining the category of technical condition of building structures is presented. A feature of the description of the state of a construction structure is the uncertainty, the sources of which can be: the inability to accurately measure the controlled quantities; inaccuracy of performing actions; impossibility of a clear description of the situation, etc. The decision about the state is connected with the solution of the problem with very fuzzy initial data, connected with approximate «linguistic» characteristics of the input parameters (defects and structural damages, strength of materials, etc.). The process of determining the category of technical condition is represented as a set of interrelated stages (subtasks) of the solution of the task. For the rapid implementation of these stages, the system of ontological analysis has integrated the corresponding subtasks in the form of a multilevel hierarchical structure (ontograph). The hierarchical principle will allow for regulating the declarative knowledge at stages and coordinating the results of the decisions of the corresponding sub-tasks, using heterogeneous information flows. The construction of formal schemes and models for solving problems characterized by insufficient information, inconsistency and low reliability in the survey of building objects allows us to perform the theory of fuzzy sets and its applications. The definition of a clear meaning of the category of technical condition involves the use of the Mamdani fuzzy logic inference model. The formation of the corresponding Mamdani knowledge bases, in which all the values of the input and output variables are given by fuzzy sets, does not cause difficulties for the expert. The results of the research were used to develop a research version of the expert system for assessing the technical condition of reinforced concrete bending elements. Approbation and verification of the program was carried out on the actually existing structures of the operated buildings and structures.