Forecasting

This package contains the simulation and methods for creating forecasts. They are split into statistical forecasts and weather forecast.

• Simulation
• Statistical Forecasting
  • Double-Seasonal Forecast
  • Day-Type Forecast
• Holt-Winters Methods
  • Choltwinters — Holt-Winters Extensions
• Weather
• Helping functions for Forecasting
  • helpers
  • dataloader

Forecasts

get_forecast(initial_time, configurations=None, code=None, forward=None)

class Forecast(initial_time, configurations=None, code=None, forward=None, forecast=True)

Bases: threading.Thread

Setup a Forecast Object. A new BaseEnvironment and new Devices will be created. Forecasting can either be ran synchronous or asynchronous (threaded):

foocast = Forecast(time.time(), forward=10*24*3600)
barcast = Forecast(time.time(), forward=2*24*3600)

#run threaded
barcast.start() 

#wait until foocast is finished, then get result
resultfoo = foocast.run().get() 

# wait until barcast is finished
while resultbar == None:
    resultbar = barcast.get()

Parameters:

• initial_time (int) – timestamp of the time, at which the forecast starts
• configurations – cached configurations, if None, retrieve from database
• code – code to be executed
• forward (int) – Time to forecast. Uses DEFAULT_FORECAST_INTERVAL if None
• forecast (boolean) – Passed to BaseEnvironment forecast.

get()

return the result of the forecast. If the mainloop is still forecasting, None is returned.

outputs a dict with:

result = {start: datetime, 
           step: stepsize, 
            end: datetime, 
        sensors: list with values per sensor (see MeasurementStorage)}

run()

run the main loop. Returns self after finishing. Results are obtained with get()

step()

execute one step of the simulation. This steps all devices, auto-optimizes if needed and store the values

store_values()

sample device values

class DemoSimulation(initial_time, configurations=None)

Bases: server.forecasting.Forecast

A Forecast, which writes the values to the database. It replaces the real devices and is used to develop and show the capabilities of ecoControl. It uses real electrical and weather values instead of forecasts, the device simulation on the other hand is the same as in Forecast.

After calling start(), the simulation will currently run at 30 steps per second (or 30x speed). This is controlled by the step_size in BaseEnvironment.

The simulation can be forwarded to a certain point by setting the forward variable in seconds > 0. It will then run at maximum speed. The simulation runs until the variable running is set to False.

Note

DemoSimulations should generally be started with start_or_get()

run()

run while running is true, call the parent step() method. This method must be called by start(), otherwise it immediately returns

start()

start the simulation in a seperate thread

classmethod start_or_get(print_visible=False)

This method starts a new demo simulation if necessary and it makes sure that only one demo simulation can run at once. This is the preferred way to start the demo simulation.

Returns:DemoSimulation or None if system not in demo mode.

store_values()

stores values in database. Overwrites parents saving method. Values are only stored every (simulated) minute

class ForecastQueue

A container, holding the running forecasts. Each forecast gets an id.

Usage:

q = ForecastQueue()
f_id = q.schedule_new(initial_time=time.time())
#... do other stuff, then retrieve forecast
result = q.get_by_id(f_id)

get_by_id(forecast_id)

get a forecast by its id. Will return None, if forecast is not completed. If the forecast is finished, the result is returned and deleted from the ForecastQueue.

schedule_new(initial_time, **kwargs)

start a new forecast and return its id. :param dict kwargs: the parameters for the Forecast

get_initialized_scenario(env, configurations)

this function returns an initialized scenario. It creates new simulated devices and connects the right devices.

The devices are restored to the latest state of the SensorValue‘s in the db, if there are no Values, a warning will be logged and the standard values are used.

Parameters:

• env – BaseEnvironment for all Devices
• configurations (list) – the device configurations, which to set in the devices. These are typically DeviceConfiguration objects.

Returns:

a namedtuple of devices, with the acronym (f.e plb), as key