Home » Agent Based Modeling (Event model)

Agent Based Modeling (Event model)

Agent-based modeling (ABM) is a class of computational methods used to simulate actions or behaviors and interactions between autonomous agents and assess the impact on the whole environment of such interactions/dynamics. Such agent-based computations/modeling are used in the simulation/representation of systems that have features of complex systems, emergence of behaviors, sociology. Agent-based modeling and multi agent system are for some aspects overlapping, but are not the same. The first is more focused on the investigation of the behavior of an entire system mostly in the domain of natural and social sciences. The second is used in the domain of technical systems and discrete systems. An ABM can be created with any platform starting from excel going to the more advance platform that exploit high level of parallelism.

AgentThere are some features that have to be satisfied in order to have an agent-based model. As in [1], the features that an agent has to provide as core are:

  • Autonomy (functions independently of other agents)
  • Self-contained (modular, it is discrete and intrinsic with a set of characteristics)
  • Interaction (social or behavioral ability)

Other properties that agents usually have are:

  • Live in an environment
  • Has explicit goals
  • Learn and adapt its behaviors
  • Can have a resource attribute (stock of a resource e.g., energy)

 

In developing the SmartCap ToolBox the aim is to stay as true as possible to the main aspects of distribution network characteristics by employing ABM. However the electrical properties of the network objects and the topological hierarchy require some “agents” to operate with less degree of freedom. Other then that the “calculation thick model” has some restrictions (at the moment) in the resolution of our data. The calculation has a resolution of 15 minutes; the speed of the model is of course flexible and there are opportunities in the future to calculate all actions at a higher resolution; for now however we use the “baseload” with a 15 minutes interval. So although a lot of agents can run free within “the frame-rate of the engine” (for example electric cars), the calculations are based on 15 minutes intervals. The 15 minutes resolution stems from the current standard, the PTU (programme time unit), presently in use in the electricity sector. Moreover, smart meter interval data also has a resolution of 15 minutes.

AutonomeObjecten