The lecture by Dr. Armen Beklaryan and Professor Andranik S. Akopov will take place at the Complexity Science Hub Vienna in the Salon.
Developing simulation models for decision-making systems for economic, social, and ecological planning
Professor Andranik S. Akopov, Russian Academy of Science, Moscow
In this talk, the main research directions in the field of simulation modelling conducted at the Central Institute of Economics and Mathematics of the Russian Academy of Science will be presented.
Firstly, we developed a series of agent-based models (ABM) of human crowd behaviour in emergency. The behaviour of evacuated agents interacting with agent-rescuers is simulated considering different situations (e.g. fires, explosions, smoke spreading, etc.) that cause effects of the ‘crowd turbulence’ and the ‘crowd сrush’. Such models are integrated with the fuzzy clustering algorithm, the parallel genetic optimisation algorism, and other computational methods in order to seek the best scenarios for an evacuation.
The second direction of our research is related to developing agent-based ecological-economic models which are intended to seeking the best trade-offs in the ecological modernisation of enterprises using a case study of the Republic of Armenia. Such models allow to seeking Pareto optimal solutions that reduce total air pollution while keeping up the positive dynamics of the industrial production.
The second group of ABM-models is intended for multi-agent simulations of air pollution dynamics in cities using a case study of Yerevan, Armenia. These models are based on using natural resources, such as agent-trees for reducing air pollution concentration. Thus, we developed decision-making systems for ecological planning on a country and city level.
The third type of our models are simulations that use system dynamics (SD) and discrete-event simulations which are developed for controlling activities of vertically integrated corporations (petroleum corporations, financial corporation, etc.). Using industrial cases developed and implemented in large transnational corporations such as LUKIOL, SBERBANK, and ROSNEFT, we will demonstrate how integrated simulation models aggregated with developed genetic optimisation algorithms can be a core decision-making systems for economic and investment planning.
Note that all these models are implemented in special simulation tools, such as the AnyLogic (www.anylogic.com) and Powersim Studio (www.powersim.com), which support different methods of simulation modelling including agent-based modelling, system dynamics, discrete-event simulation, etc. Most of these models are integrated with artificial intelligence methods (AI) implemented using C++ and Java, such as the suggested multi-agent genetic optimisation algorithm for multi-objective optimisation (MAGAMO), the developed Fuzzy Clustering Algorithm for Crowd behaviour, etc. Besides, the system integration of developed simulations with databases (MS SQL Server), Web Applications and Services, and GIS will also be demonstrated.