Intelligent methods for evaluating the state of hierarchical systems

This section of the study proposes intelligent methods for assessing the state of hierarchical systems.

During the research, the authors:
– conducted an analysis of knowledge representation models, substantiating the advantages of using production knowledge representation in expert systems. The study outlines key concepts of fuzzy expert systems and formulates a formal task for accelerating decision-making in the rule base of a fuzzy expert system;
– developed a methodology for assessment and prediction using fuzzy cognitive maps.

The novelty of the proposed methodology lies in:
– considering a corrective coefficient for the degree of uncertainty regarding the state of the object;
– adding a corrective coefficient for data noise resulting from distortions in object state information;
– reducing computational costs when evaluating object states;
– creating a multilevel and interconnected description of hierarchical objects;
– adjusting the object description due to changes in its current state using a genetic algorithm;
– enabling calculations with input data of different natures and measurement units.

The research includes the development of a visualization method for hierarchical system states. The novelty of this method lies in:
– creating a visual, multilevel, and interconnected description of the hierarchical system;
– enhancing decision-making efficiency in assessing the hierarchical system’s state;
– addressing the issue of global and local extrema when evaluating the state of hierarchical systems;
– combining graphical and numerical representations of the monitored parameters of the hierarchical system’s state;
– avoiding loop formation during real-time visualization of hierarchical system states.

The study also develops a method for evaluating complex hierarchical systems based on an improved particle swarm optimization (PSO). This evaluation method combines PSO with coordinate averaging and its modification by employing multiple particle swarms and integrating the Hooke-Jeeves procedure and appropriate corrective coefficients. The novelty of this method lies in:
– creating a multilevel and interconnected description of real-time complex hierarchical systems;
– enhancing decision-making efficiency for real-time hierarchical systems assessment;
– resolving global and local extrema issues during real-time hierarchical system state evaluation;
– enabling directed searches by multiple swarm particles in a specific direction;
– considering the degree of uncertainty;
– allowing for repeated analysis of complex real-time hierarchical systems.