# US 2023/0273574 A1 · Autonomous and semantic optimization approach for real-time performance management in a built environment

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0273574 A1
- Document ID: US20230273574A1
- Status: Published application
- Kind: A1
- Application number: 18/144,820
- Filed: 2023-05-08
- Published: 2023-08-31
- Earliest claimed: 2019-07-16
- Assignees: University College Cardiff Consultants Limited (Uc3)
- Applicants: University College Cardiff Consultants Limited (Uc3)
- Inventors: Yacine REZGUI; Thomas BEACH; Wanqing ZHAO; Muhammad AHMAD; Ioan PETRI; Jean-Laurent HIPPOLYTE; Shaun HOWELL
- Web page: https://patentgrep.com/patent/US20230273574A1

## CPC

- G05B 13/041: {in which a variable is automatically adjusted to optimise the performance}
- F24F 11/50: characterised by user interfaces or communication
- G05B 13/0265: {the criterion being a learning criterion}
- G06F 40/30: Semantic analysis
- G06N 3/086: using evolutionary algorithms, e.g. genetic algorithms or genetic programming
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06N 5/027: {Frames}
- G06N 20/00: Machine learning
- G06N 3/126: Evolutionary algorithms, e.g. genetic algorithms or genetic programming
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- F24F: AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING (removing dirt or fumes from areas where they are produced B08B15/00; vertical ducts for carrying away waste gases from buildings E04F17/02; tops for chimneys or ventilating shafts, terminals for flues F23L17/02)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Certain aspects of the present disclosure provide techniques for semantically contextualizing structured and unstructured data, extracting and making sense of domain knowledge, and performing semantic-driven optimization to provide generic, scalable, autonomous and real-time performance management within a built environment domain.

## Claims

### Claim 1

A method for semantically-driven optimization, comprising:
receiving, from a user, a request for an optimization event;
obtaining one or more domain and user requirements associated with the optimization event for a semantic domain model;
determining a prediction model and one or more prediction model parameters associated with the optimization event;
determining a simulation model and one or more simulation model parameters associated with the optimization event;
generating a generalized optimization problem based on the optimization event and the one or more domain and user requirements;
determining an optimization model based on the generalized optimization problem;
generating an optimization output from the optimization model; and
providing the optimization output to the user.

### Claim 2 (depends on 1)

The method of claim 1, further comprising modelling the one or more domain and user requirements associated with the optimization event based on a generalized characteristics representation.

### Claim 3 (depends on 1)

The method of claim 1, wherein determining the prediction model and the one or more prediction model parameters based on the one or more domain and user requirements, comprises:
identifying the prediction model from a plurality of prediction models associated with the optimization event and the one or more domain and user requirements;
determining the one or more prediction model parameters based on the prediction model and the one or more domain and user requirements;
determining historical data associated with the one or more domain and user requirements; and
providing the prediction model and the prediction model parameters to the optimization model.

### Claim 4 (depends on 1)

The method of claim 1, wherein determining the prediction model and the one or more prediction model parameters based on the one or more domain and user requirements, comprises:
determining no prediction model of a plurality of prediction models is associated with the optimization event and the one or more user and domain requirements,
constructing a new predictive model semantically based on historical data, comprising:
obtaining the historical data associated with the one or more domain and user requirements associated with the optimization event and the optimization event;
determining, via a variable selection service the one or more prediction model parameters based on the one or more domain and user requirements; and
training the new predictive model based on the one or more prediction model parameters and the historical data; and
providing the new prediction model and the one or more prediction model parameters to the optimization model.

### Claim 5 (depends on 1)

The method of claim 1, wherein determining the simulation model and the one or more simulation model parameters associated with the optimization event, comprises:
identifying the simulation model from a plurality of simulation models associated with the optimization event and the one or more domain and user requirements;
determining the one or more simulation model parameters based on the simulation model and the one or more domain and user requirements;
determining simulation data associated with the one or more domain and user requirements; and
providing the simulation model and the simulation model parameters to the optimization model.

### Claim 6 (depends on 1)

The method of claim 1, wherein determining the simulation model and the one or more simulation model parameters based on the one or more domain and user requirements, comprises:
determining no simulation model of a plurality of simulation models is associated with the optimization event and the one or more user and domain requirements, constructing a new simulation model semantically based on simulation data, comprising:
obtaining the simulation data associated with the one or more domain and user requirements associated with the optimization event and the optimization event;
determining, via a variable selection service the one or more simulation model parameters based on the one or more domain and user requirements; and
training the new simulation model based on the one or more simulation model parameters and the simulation data; and
providing the new simulation model and the one or more simulation model parameters to the optimization model.

### Claim 7 (depends on 1)

The method of claim 1, wherein generating the optimized output based on the optimization model, comprises;
executing the prediction model with one or more intermediate optimized solutions generated by the optimization model and prediction data;
executing the simulation model with the one or more intermediate optimized solutions generated by the optimization model and simulation data; and
obtaining the optimized output from the optimization model based on the executed prediction model and the executed simulation model.

### Claim 8 (depends on 1)

The method of claim 1, wherein the request for the optimization event comprises a structured natural language statement.

### Claim 9

A method for semantically-driven optimization, comprising:
receiving, from a user, a request for an optimization event;
obtaining one or more domain and user requirements associated with the optimization event for a semantic domain model;
determining a prediction model and one or more prediction model parameters based on the one or more domain and user requirements;
generating a generalized optimization problem based on the optimization event and the one or more domain and user requirements;
determining an optimization model based on the generalized optimization problem;
generating an optimization output from the optimization model; and
providing the optimization output to the user.

### Claim 10 (depends on 9)

The method of claim 9, further comprising: modelling the one or more domain and user requirements based on a generalized characteristics representation.

### Claim 11 (depends on 9)

The method of claim 9, wherein determining the prediction model and the one or more prediction model parameters based on the one or more domain and user requirements, comprises:
determining no prediction model of a plurality of prediction models is associated with the optimization event and the one or more user and domain requirements,
constructing a new predictive model semantically based on historical data, comprising:
determining, via a variable selection service, the one or more prediction model parameters based on the one or more domain and user requirements; and
training the new predictive model based on the historical data and simulated data; and
providing the new prediction model to the optimization model.

### Claim 12 (depends on 11)

The method of claim 11, wherein determining, via the variable selection service, the one or more prediction model parameters based on the one or more domain and user requirements, comprises:
determining one or more variables and constraints based on the one or more domain and user requirements;
determining input variables from the one or more variables and constraints based on a semantic relationship between the one or more variables and constraints and output variables for the prediction model; and
selecting, the one or more prediction model parameters from the input variables via the variable selection service.

### Claim 13 (depends on 9)

The method of claim 9, wherein generating the optimized output based on the optimization model, comprises;
executing the prediction model with one or more intermediate optimized solutions generated by the optimization model and prediction data; and
obtaining the optimized output from the optimization model based on the executed prediction model.

### Claim 14

A processing system for a semantically-driven optimization, comprising: a memory comprising computer-executable instructions; and a processor configured to execute the computer-executable instructions and cause the processing system to:
receive, from a user, a request for an optimization event;
obtain one or more domain and user requirements associated with the optimization event for a semantic domain model;
determine a prediction model and one or more prediction model parameters associated with the optimization event;
determine a simulation model and one or more simulation model parameters associated with the optimization event;
generate a generalized optimization problem based on the optimization event and the one or more domain and user requirements;
determine an optimization model based on the generalized optimization problem;
generate an optimization output from the optimization model; and
provide the optimization output to the user.

### Claim 15 (depends on 14)

The processing system of claim 14, wherein the processing system is further configured to model the one or more domain and user requirements associated with the optimization event based on a generalized characteristics representation.

### Claim 16 (depends on 14)

The processing system of claim 14, wherein in order to determine the prediction model and the one or more prediction model parameters based on the one or more domain and user requirements, the processing system is configured to:
identify the prediction model from a plurality of prediction models associated with the optimization event and the one or more domain and user requirements;
determine the one or more prediction model parameters based on the prediction model and the one or more domain and user requirements;
determine historical data associated with the one or more domain and user requirements; and
provide the prediction model and the prediction model parameters to the optimization model.

### Claim 17 (depends on 14)

The processing system of claim 14, wherein in order to determine the prediction model and the one or more prediction model parameters based on the one or more domain and user requirements, the processing system is configured to:
determine no prediction model of a plurality of prediction models is associated with the optimization event and the one or more user and domain requirements,
construct a new predictive model semantically based on historical data, comprising:
obtain the historical data associated with the one or more domain and user requirements associated with the optimization event and the optimization event;
determine, via a variable selection service the one or more prediction model parameters based on the one or more domain and user requirements; and
train the new predictive model based on the one or more prediction model parameters and the historical data; and
provide the new prediction model and the one or more prediction model parameters to the optimization model.

### Claim 18 (depends on 14)

The processing system of claim 14, wherein in order to determine simulation model and the one or more simulation model parameters associated with the optimization event, the processing system is configured to:
identify the simulation model from a plurality of simulation models associated with the optimization event and the one or more domain and user requirements;
determine the one or more simulation model parameters based on the simulation model and the one or more domain and user requirements;
determine simulation data associated with the one or more domain and user requirements; and
provide the simulation model and the simulation model parameters to the optimization model.

### Claim 19 (depends on 14)

The processing system of claim 14, wherein in order to determine the simulation model and the one or more simulation model parameters based on the one or more domain and user requirements, the processing system is configured to:
determine no simulation model of a plurality of simulation models is associated with the optimization event and the one or more user and domain requirements,
construct a new simulation model semantically based on simulation data, comprising:
obtain the simulation data associated with the one or more domain and user requirements associated with the optimization event and the optimization event;
determine, via a variable selection service the one or more simulation model parameters based on the one or more domain and user requirements; and
train the new simulation model based on the one or more simulation model parameters and the simulation data; and
provide the new simulation model and the one or more simulation model parameters to the optimization model.

### Claim 20 (depends on 15)

The processing system of claim 15, wherein in order to generate the optimized output based on the optimization model, the processing system is configured to;
execute the prediction model with one or more intermediate optimized solutions generated by the optimization model and prediction data;
execute the simulation model with the one or more intermediate optimized solutions generated by the optimization model and simulation data; and
obtain the optimized output from the optimization model based on the executed prediction model and the executed simulation model.
