# US 2025/0085673 A1 · Utility usage prediction and optimization

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

- Number: US 2025/0085673 A1
- Document ID: US20250085673A1
- Status: Published application
- Kind: A1
- Application number: 18/465,413
- Filed: 2023-09-12
- Published: 2025-03-13
- Earliest claimed: 2023-09-12
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Srinath Madasu
- Web page: https://patentgrep.com/patent/US20250085673A1

## CPC

- G05B 13/027: {using neural networks only}
- G06Q 50/06: Energy or water supply
- G06Q 2220/10: Usage protection of distributed data files
- 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)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Techniques are provided for utility usage prediction and optimization. In one embodiment, the techniques involve receiving user data, receiving utility data, generating, via a trained physics-informed neural network (PINN), a utility usage prediction based on the utility data, generating a utility plan based on the user data and the utility usage prediction, wherein the utility plan includes limits or restrictions of a utility usage, and controlling the utility usage based on the utility plan.

## Claims

### Claim 1

A method comprising:
receiving user data;
receiving utility data;
generating, via a trained physics-informed neural network (PINN), a utility usage prediction based on the utility data;
generating a utility plan based on the user data and the utility usage prediction, wherein the utility plan includes limits or restrictions of a utility usage; and
controlling the utility usage based on the utility plan.

### Claim 2 (depends on 1)

The method of claim 1, wherein the user data includes at least one of: a utility budget, a utility usage analytics request, a utility usage prediction request, or an optimized utility plan request;
wherein the utility data includes data from a utility system, a weather system, or a metering system;
wherein the limits or restrictions of the utility plan indicate the utility usage during a time window or during a given time of a day; and
wherein the limits or restrictions of the utility plan indicate the utility usage in compliance with the utility budget.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
transferring the utility plan to a user; and
recording, via a distributed ledger technology, user payments associated with the utility plan.

### Claim 4 (depends on 1)

The method of claim 1, wherein training the trained PINN comprises:
training a first local PINN based on utility data of a first local environment;
training a second local PINN based on utility data of a second local environment;
training a global PINN based on the first local PINN and the second local PINN; and
determining the trained PINN based on one of: the first local PINN, the second local PINN, or the global PINN, wherein the trained PINN is a federated PINN.

### Claim 5 (depends on 4)

The method of claim 4, wherein upon determining that the utility data includes incomplete data or extrapolated data, the trained PINN is determined based on the first local PINN or the second local PINN.

### Claim 6 (depends on 4)

The method of claim 4, wherein upon determining that the utility data does not include incomplete data or extrapolated data, the trained PINN is determined based on the global PINN.

### Claim 7 (depends on 4)

The method of claim 4, wherein the global PINN is trained based on a weighted average of local PINNs such that GlobalLOSS=PhysicsPDE+Ei=1Nwī*LocalLOSS(i), where GlobalLOSS represents a loss function of the global PINN, PhysiCSPDE represents a physics partial differential equation, wī represents a weighted average of an ith local PINN, and LocalLOSS (i) represents a loss function of an ith PINN.

### Claim 8

A system, comprising:
a processor; and
memory or storage comprising an algorithm or computer instructions, which when executed by the processor, performs an operation comprising:
receiving user data;
receiving utility data;
generating, via a trained physics-informed neural network (PINN), a utility usage prediction based on the utility data;
generating a utility plan based on the user data and the utility usage prediction, wherein the utility plan includes limits or restrictions of a utility usage; and
controlling the utility usage based on the utility plan.

### Claim 9 (depends on 8)

The system of claim 8, wherein the user data includes at least one of: a utility budget, a utility usage analytics request, a utility usage prediction request, or an optimized utility plan request;
wherein the utility data includes data from a utility system, a weather system, or a metering system;
wherein the limits or restrictions of the utility plan indicate the utility usage during a time window or during a given time of a day; and
wherein the limits or restrictions of the utility plan indicate the utility usage in compliance with the utility budget.

### Claim 10 (depends on 8)

The system of claim 8, the operation further comprising:
transferring the utility plan to a user; and
recording, via a distributed ledger technology, user payments associated with the utility plan.

### Claim 11 (depends on 8)

The system of claim 8, wherein training the trained PINN comprises:
training a first local PINN based on utility data of a first local environment;
training a second local PINN based on utility data of a second local environment;
training a global PINN based on the first local PINN and the second local PINN; and
determining the trained PINN based on one of: the first local PINN, the second local PINN, or the global PINN, wherein the trained PINN is a federated PINN.

### Claim 12 (depends on 11)

The system of claim 11, wherein upon determining that the utility data includes incomplete data or extrapolated data, the trained PINN is determined based on the first local PINN or the second local PINN.

### Claim 13 (depends on 11)

The system of claim 11, wherein upon determining that the utility data does not include incomplete data or extrapolated data, the trained PINN is determined based on the global PINN.

### Claim 14 (depends on 11)

The system of claim 11, wherein the global PINN is trained based on a weighted average of local PINNs such that GlobalLOSS=PhysicsPDE+Ei=1Nwī*LocalLOSS(i), where GlobalLOSS represents a loss function of the global PINN, PhysicSPDE represents a physics partial differential equation, wi represents a weighted average of an ith local PINN, and LocalLOSS(i) represents a loss function of an ith PINN.

### Claim 15

A computer-readable storage medium having a computer-readable program code embodied therewith, the computer-readable program code executable by one or more computer processors to perform an operation comprising:
receiving user data;
receiving utility data;
generating, via a trained physics-informed neural network (PINN), a utility usage prediction based on the utility data;
generating a utility plan based on the user data and the utility usage prediction, wherein the utility plan includes limits or restrictions of a utility usage; and
controlling the utility usage based on the utility plan.

### Claim 16 (depends on 15)

The computer-readable storage medium of claim 15, wherein the user data includes at least one of: a utility budget, a utility usage analytics request, a utility usage prediction request, or an optimized utility plan request;
wherein the utility data includes data from a utility system, a weather system, or a metering system;
wherein the limits or restrictions of the utility plan indicate the utility usage during a time window or during a given time of a day; and
wherein the limits or restrictions of the utility plan indicate the utility usage in compliance with the utility budget.

### Claim 17 (depends on 15)

The computer-readable storage medium of claim 15, wherein training the trained PINN comprises:
training a first local PINN based on utility data of a first local environment;
training a second local PINN based on utility data of a second local environment;
training a global PINN based on the first local PINN and the second local PINN; and
determining the trained PINN based on one of: the first local PINN, the second local PINN, or the global PINN, wherein the trained PINN is a federated PINN.

### Claim 18 (depends on 17)

The computer-readable storage medium of claim 17, wherein upon determining that the utility data includes incomplete data or extrapolated data, the trained PINN is determined based on the first local PINN or the second local PINN.

### Claim 19 (depends on 17)

The computer-readable storage medium of claim 17, wherein upon determining that the utility data does not include incomplete data or extrapolated data, the trained PINN is determined based on the global PINN.

### Claim 20 (depends on 17)

The computer-readable storage medium of claim 17, wherein the global PINN is trained based on a weighted average of local PINNs such that GlobalLOSS=PhysicsPDE+Ei=1Nwī*LocalLOSS(i), where GlobalLOSS represents a loss function of the global PINN, PhysiCSPDE represents a physics partial differential equation, wī represents a weighted average of an ith local PINN, and LocalLOSS(i) represents a loss function of an ith PINN.