neural_policy_controller

Modules

neural_policy_controller_system module

class NeuralPolicyControllerSystem(input_size=None, output_size=None, input_output_schema=None, policy_model=None, **kwargs)

Bases: System

Neural policy controller for RL-based building control.

This class implements a neural network-based controller that uses reinforcement learning techniques to optimize building energy performance. The controller takes the current state of the building as input and outputs control signals based on a trained policy.

Features:

• The controller is based on a neural network model that takes as input the current state of the building and outputs the control signal

• The neural network model is trained using reinforcement learning techniques to optimize building energy performance

• The input and output of the controller is defined by a JSON schema that contains the keys and types of the input and output signals

• The neural policy is initialized at instantiation and the weights are updated manually by the user, typically through a training process

Parameters:

• input_size (int) – Size of the input state vector

• output_size (int) – Size of the output control vector

• input_output_schema (dict) – JSON schema defining input/output structure and ranges

• policy_model (nn.Module, optional) – Pre-trained neural network policy. If None, a default architecture is created.

• **kwargs – Additional keyword arguments passed to the parent System class

denormalize_output_data(data)

Denormalize the output data using the schema. Inputs: data (numpy array or tensor of shape (output_size,)) Outputs: denormalized data (numpy array) The min and max values are stored in the input_output_schema[“output”] dictionary.

do_step(secondTime, dateTime, step_size, stepIndex)

Do a single step of the system.

Parameters:

• secondTime (float) – The current time in seconds.

• dateTime (datetime.datetime) – The current date and time.

• step_size (int) – The step size of the simulation in seconds.

• stepIndex (int) – The current step index.

Return type:

None

initialize(start_time, end_time, step_size, simulator)

Initialize the system.

Parameters:

• start_time (datetime.datetime) – The start time of the simulation.

• end_time (datetime.datetime) – The end time of the simulation.

• step_size (int) – The step size of the simulation in seconds.

• simulator (core.Simulator) – The simulator.

Return type:

None

load_policy_model(policy_path)

normalize_input_data(data)

select_action(state)

validate_schema(data)

property config