Method and controller for controlling a chiller plant for a building and chiller plant

1. A computer implemented method for controlling a chiller plant for a building, comprising:

a) receiving, as first training data, environmental data of an environment of the building and cooling load demand data,

b) using the first training data for training a first machine learning model to predict a cooling load demand from environmental data,

c) receiving, as second training data, control signals for the chiller plant and cooling power data resulting from applying the control signals to the chiller plant,

d) using the second training data for training a second machine learning model to predict a cooling power from control signals,

e) receiving actual environmental data and predicting a resulting cooling load demand by the trained first machine learning model,

f) generating a plurality of candidate control signals for the chiller plant, sending the plurality of generated candidate control signals to the trained second machine learning model, and predicting a resulting cooling power by the trained second machine learning model,

g) receiving the generated plurality of candidate control signals, resulting cooling load demand, and resulting cooling power and selecting, from the plurality of generated candidate control signals, applicable control signals for which the resulting cooling power fulfills the resulting cooling load demand, and

h) outputting the applicable control signals for controlling the chiller plant.

2. The method as claimed in claim 1 , further comprising:

receiving, as second training data, power consumption data resulting from applying the control signals to the chiller plant,

using the power consumption data for training the second machine learning model to predict a power consumption of the chiller plant from control signals,

predicting from the candidate control signals by the trained second machine learning model a resulting power consumption, and

selecting from the candidate control signals applicable control signals for which the predicted power consumption is lower than a predicted power consumption resulting from other candidate control signals.

3. The method as claimed in claim 1 , further comprising:

receiving building data regarding a structure or status of at least one of the building and occupancy data regarding an occupancy of the building, and

taking the at least one of the building data and the occupancy data into account in the at least one of the training of the first machine learning model and in the training of the second machine learning model.

4. The method as claimed in claim 1 , wherein

several components of the chiller plant with different inner dynamics are jointly modeled by the second machine learning model in an at least partially component-agnostic manner.

5. The method as claimed claim 1 , wherein

the second machine learning model comprises a partial machine learning model specifically adapted for modeling a particular component of the chiller plant.

6. The method as claimed in claim 1 , further comprising:

generating a recommendation data record from the applicable control signals, and

outputting the recommendation data record via a user interface.

7. The method as claimed in claim 1 , wherein

at least one of the first machine learning model and the second machine learning model comprises at least one of an artificial neural network, a recurrent neural network, a convolutional neural network, a Bayesian network, an autoencoder, a deep learning architecture, a reinforcement learning model, a support vector machine, a data driven trainable regression model, a k-nearest neighbor classifier, a physical model and a decision tree.

8. The method as claimed in claim 1 , comprising:

measuring further at least one of first training data and further second training data during operation of the chiller plant, and

further training at least one of the trained first machine learning model and the trained second machine learning model during the operation of the chiller plant by at least one of the further first training data and the further second training data.

9. The method as claimed in claim 1 , wherein

the applicable control signals are determined by at least one of a particle swarm optimization method, a genetic algorithm and a gradient decent method.

10. A controller for controlling a chiller plant for a building, adapted to perform the method according to claim 1 .

11. A chiller plant with the controller according to claim 10 .

12. A computer program product, comprising a computer readable hardware storage device having computer readable program code stored therein, said program code executable by a processor of a computer system to implement a method for controlling a chiller plant for a building, the method including:

a) receiving, as first training data, environmental data of an environment of the building and cooling load demand data,

b) using the first training data for training a first machine learning model to predict a cooling load demand from environmental data,

c) receiving, as second training data, control signals for the chiller plant and cooling power data resulting from applying the control signals to the chiller plant,

d) using the second training data for training a second machine learning model to predict a cooling power from control signals,

e) receiving actual environmental data and predicting a resulting cooling load demand by the trained first machine learning model,

f) generating a plurality of candidate control signals for the chiller plant, sending the plurality of generated candidate control signals to the trained second machine learning model, and predicting a resulting cooling power by the trained second machine learning model,

g) receiving the plurality of generated candidate control signals, resulting cooling load demand, and resulting cooling power and selecting, from the plurality of generated candidate control signals, applicable control signals for which the resulting cooling power fulfills the resulting cooling load demand, and

h) outputting the applicable control signals for controlling the chiller plant.

13. A non-transient computer readable storage medium storing a computer program product according to claim 12 .