Optimization control technology for building energy conservation

1. A control system for controlling environmental conditions within a structure having at least one environmental control device, comprising: a simulation process implemented by at least one of a computer or a processor, for carrying out a computerized simulation process, to generate and store a simulation model based on conditions associated with the structure and to provide simulation results as data;

a neural network model, implemented using at least one of the computer or the processor, programmed to implement a neural network, the neural network having an input layer coupled to receive sensor data from the structure and having an output layer coupled to supply control signals to the at least one environmental control device, the neural network model trains the neural network using the simulation model;

an optimization computer running a particle swarm optimization algorithm to receive the simulation results and perform particle swarm optimization to ascertain a control parameter regime for controlling the at least one environmental control device and supplying said control parameter regime to the neural network model;

the neural network model being further programmed to use the control parameter regime supplied by the particle swarm optimization algorithm to further train the neural network; and

wherein the neural network model is trained using:

static conditions indicating how the structure responds to thermal conditions, and

dynamic, time-varying conditions to predict different responsive behaviors under the dynamic, time-varying conditions, and where the different, responsive behaviors are used to predict energy consumption levels under the dynamic, time-varying conditions.

2. The control system of claim 1 wherein the simulation process, implemented by the at least one of the computer or the processor, implements a compound simulation model comprising:

a thermal load model, a structure envelope model, a heating, ventilation and air conditioning (HVAC) model and an occupants model.

3. The control system of claim 1 wherein the neural network model implements a recurrent neural network to predict individual output control parameters.

4. The control system of claim 1 wherein the neural network model implements a convolutional neural network to learn overall shape of at least one performance curve of the structure.

5. The control system of claim 1 wherein the neural network model is trained using the simulation model to which time-varying conditions have been applied to generate dynamic information.

6. The control system of claim 1 wherein at least two of the simulation process, neural network model and particle swarm optimization algorithm are both implemented using the at least one of the processor or the computer, according to different programmatic instructions.

7. The control system of claim 1 wherein the particle swarm optimization operates on parameters associated with heating, ventilation and air conditional (HVAC) equipment disposed at said structure.

8. The control system of claim 1 further comprising a software agent forming a deep reinforcement learning process that responds to an environment associated with the structure and functions to train the neural network model so that the neural network training is updated to account for changes in the environment.

9. A control system for controlling environmental conditions within a structure having at least one environmental control device, comprising:

a simulation process, implemented on at least one of a processor or computer, for carrying out a computerized simulation process, to generate and store a simulation model based on conditions associated with the structure and to provide simulation results as data;

a neural network model, implemented by at least one of the processor or the computer, and programmed to implement a neural network, the neural network having an input layer coupled to receive sensor data from the structure and having an output layer coupled to supply control signals to the at least one environmental control device, the neural network model being further programmed to train the neural network using the simulation model;

an optimization computer running a particle swarm optimization algorithm to receive the simulation results and perform particle swarm optimization to ascertain a control parameter regime for controlling the at least one environmental control device and supplying said control parameter regime to the neural network model; and

a software agent forming a deep reinforcement learning process that responds to an environment associated with the structure and functions to train the neural network model so that the neural network training is updated to account for changes in the environment

the neural network model being further programmed to use the control parameter regime supplied by the particle swarm optimization algorithm to further train the neural network using at least one of;

static conditions indicating how the structure responds to thermal conditions, and

dynamic, time-varying conditions to predict different responsive behaviors that influence energy consumption levels;

wherein the neural network drives a plurality of settings of coupled heating, ventilation and air conditional (HVAC) equipment,

wherein the neural network model implements a convolutional neural network to learn overall shape of at least one performance curve of the structure.