Filters API

The observations are the variables, parameters, meters, and others that the agent is capable to sense. The following code-block present an example of an observation configuration with the help of ObservationSpec to show how is the correct way of set this agent property.

from eprllib.Agent.AgentSpec import ObservationSpec

observation = ObservationSpec(
    variables = [
        ("Site Outdoor Air Drybulb Temperature", "Environment"),
        ("Site Wind Speed", "Environment"),
        ("Site Outdoor Air Relative Humidity", "Environment"),
        ("Zone Mean Air Temperature", "Thermal Zone: Room1"),
        ("Zone Air Relative Humidity", "Thermal Zone: Room1"),
        ("Zone People Occupant Count", "Thermal Zone: Room1"),
        ("Zone Thermal Comfort CEN 15251 Adaptive Model Category I Status", "Room1 Occupancy"),
        ("Zone Thermal Comfort CEN 15251 Adaptive Model Category II Status", "Room1 Occupancy"),
        ("Zone Thermal Comfort CEN 15251 Adaptive Model Category III Status", "Room1 Occupancy"),
    ],
    internal_variables = [
        ("Zone Air Volume", "Thermal Zone: Room1"),
        ("Zone Floor Area", "Thermal Zone: Room1"),
    ],
    simulation_parameters = {
        'today_weather_horizontal_ir_at_time': True,
    },
    meters = [
        "Electricity:Building",
        "Heating:DistrictHeatingWater",
        "Cooling:DistrictCooling",
    ],
    use_one_day_weather_prediction = True,
    prediction_hours = 3,
    prediction_variables = {
        'outdoor_dry_bulb': True,
    },
    use_actuator_state = True,
    other_obs = {
        "WWR-North": ((2.39-0.01)*(2.28-1.48))/((3.08)*(2.4)), # WWR: Window-Wall Ratio
        "WWR-South": ((2.39-0.01)*(2.28-1.48))/((3.08)*(2.4)),
        "WWR-East": 0.0,
        "WWR-West": 0.0,
    }
),

Note that observation_fn is not part of the observation capabilities of the agent. The observation function is defined in the environment because it defines the relationship between agents as well.