Agent Overview
The Agent represents an energy system participant, such as a household, commercial entity, or industrial consumer, that interacts with markets and the grid. Each agent follows a structured workflow to make decisions, optimize its energy usage, and engage in trading.
What Does an Agent Do?
Each agent in HAMLET: - Retrieves grid data to understand network constraints. - Obtains forecasts to predict future energy consumption, production, and prices. - Executes control strategies via the Energy Management System (EMS). - Trades energy by submitting bids and offers to the market.
Agent Execution Workflow
Each simulation step follows a structured sequence:
Grid Data Retrieval - The agent gathers grid-related data to assess constraints and available capacity.
Forecasting - Agents predict their future energy needs and availability using forecasting models. - Forecasts can be based on historical data, weather predictions, or machine learning techniques.
Control Strategy Execution - The Energy Management System (EMS) defines how the agent manages its energy usage. - The EMS can follow:
Rule-based strategies
Optimization models (e.g., linear programming)
Reinforcement learning-based decisions
Market Participation - Based on its forecast and EMS, the agent submits bids and offers to the market. - Market clearing determines how much energy is bought or sold.
Agent Structure
Each agent consists of the following components:
Agent Type - The category of the agent (e.g., single-family home, multi-family home, industry). - Defines the agent’s properties such as load profiles, generation capacity, and flexibility.
Energy Management System (EMS) - The EMS defines how the agent interacts with energy markets and storage systems. - Determines when to store, consume, or trade energy.
Trading Strategy - Defines how the agent participates in energy trading. - Strategies include:
Retailer-based trading: Agents buy and sell at retailer prices.
Market-driven strategies: Agents bid dynamically based on forecasts.
Zero Intelligence (ZI) models: Randomized trading behaviors.
Grid Interaction - Ensures that the agent’s transactions respect grid constraints. - If grid limitations exist, the agent may adjust its trading behavior.
Extending Agent Behavior
HAMLET allows customization of agent behavior:
Custom Forecasting Models - Users can integrate different forecasting techniques, from simple averages to deep learning models.
Advanced EMS Control - The EMS can be customized to include complex decision-making mechanisms.
New Trading Strategies - Users can define new trading mechanisms beyond the default strategies.
By modeling agents with autonomous decision-making capabilities, HAMLET provides a powerful simulation environment for decentralized energy markets.