A Neurograph as a Model to Support Control Over the Comprehensive Objects Safety for BIM Technologies

All-Russian research-to-practice conference "Ecology and safety in the technosphere: current problems and solutions"

工业技术（总论）

Korneev, N.V.^1

Togliatti State University, 14, Belorusskaya St., Togliatti

445667, Russia^1

关键词: Adaptive fuzzy neural network;    Automated control systems;    Component-based models;    Corresponding state;    Flame propagation;    Models and algorithms;    Neural network model;    Operating condition;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/224/1/012021/pdf
DOI  :  10.1088/1755-1315/224/1/012021

学科分类：工业工程学

来源: IOP

【 摘 要 】

Control over the comprehensive security of facilities requires scientific studying models and algorithms of control support, in particular, development of the methods and algorithms of forming component models of intruders for antiterrorist and anti-criminal protection of facilities and ensuring fire safety. Building a component-based automated control system to support comprehensive security of facilities with the aim of implementing particular BIM technologies improves security, as it allows studying in detail and assessing all the risks that may occur in the operating conditions. The article describes the developed algorithm of solving the problem of supporting control over comprehensive security of facilities by automating the process of creating component-based models of intruders in general terms for the cases where the finite number of states of the controlled facility at each moment is known or unknown. The algorithm is based on the model of the facility proposed by the author, which is similar to a graphic chart, where the vertices represent the neural network that models the corresponding state, the barriers of the facility in real time, and the edges - allowed paths of transition from one state to another, supplemented by tuples , where x is the input vector, and d is the corresponding expected output vector of the network. An adaptive fuzzy neural network with a fuzzy-controller is used as the neural network. The resulting model is called a neurographic model of the facility, and the graphic chart in its base - a neurograph. The possibility is shown of amending the neurograph with any external effects described in a similar way by a relevant external neural network, on the example of neural network modeling two-dimensional flame propagation in an enclosed space with the use of the Kuramoto-Sivashinski equation. A generalized condition of transition from vertex to vertex for any arbitrarily complex neurograph has been formulated. The possibility of its implementation has been shown on an example of facility formalization with an intruder.