# US 12,224,585 B2 · Anomaly detection in energy systems

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

- Number: US 12,224,585 B2
- Document ID: US12224585B2
- Status: Granted
- Kind: B2
- Application number: 17/753,766
- Filed: 2019-09-12
- Published: 2022-11-24
- Granted: 2025-02-11
- Earliest claimed: 2019-09-12
- Assignees: The University of Hong Kong
- Applicants: The University of Hong Kong
- Inventors: Wing Tat Pong; Ke Zhu; Qi Xu
- Web page: https://patentgrep.com/patent/US12224585B2

## CPC

- H02J 3/0012: characterised by the contingency detection means in AC networks, e.g. using phasor measurement units [PMU], synchrophasors or contingency analysis
- G01R 13/00: Arrangements for displaying electric variables or waveforms
- G01R 19/2513: {Arrangements for monitoring electric power systems, e.g. power lines or loads; Logging}
- H02J 13/12: Monitoring network conditions, e.g. electrical magnitudes or operational status
- H02J: ELECTRIC POWER NETWORKS; CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- G01R: MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES (indicating correct tuning of resonant circuits H03J3/12)

## Abstract

A method and system are provided for anomaly detection in energy systems. Non-contact sensing of an energy system based on electric and magnetic fields uses non-contact electric- and magnetic-field sensors to produce electric- and magnetic-field signals. The electric and magnetic field signals are filtered to remove noise. Features are extracted and normalized from the magnetic and electric field signals to characterize parameters of each signal. Density-based spatial clustering of extracted features is performed using a selected minimum number of points required to form a cluster and a parameter indicating the distance within which data are considered to fall within the cluster. An anomaly is determined from data point(s) that do not fall within the cluster formed by data points in normal operation. The density-based spatial clustering of extracted features may be performed using a Density-Based Spatial Clustering of Application with Noise (DBSCAN) algorithm. Features may be extracted using Fourier analysis.

## Claims

### Claim 1

An anomaly detector for energy systems, the anomaly detector comprising:
an array of non-contact magnetic field sensors configured to sense a magnetic field from an energy system to which it is disposed adjacent and to produce magnetic field signals indicating operating currents in the energy system;
an array of non-contact electric field sensors configured to sense an electric field from the energy system to which it is disposed adjacent and to produce electric field signals indicating operating voltages in the energy system;
an analog-to-digital converter in operable communication with the array of non-contact magnetic field sensors and the array of non-contact electric field sensors, the analog-to-digital converter being configured to convert the magnetic field signals and the electric field signals into digital signals;
a filter to remove noise from the magnetic field signals and the electric field signals;
a feature extractor to extract features from the current and voltage signals to characterize measurement parameters of each signal;
a data processor configured to perform data clustering of extracted features; and
an anomaly indicator identifying an anomaly from one or more data points that do not fall within a cluster.

### Claim 2 (depends on 1)

The anomaly detector for energy systems according to claim 1, wherein each magnetic field sensor is an anisotropic magnetoresistive sensor, a tunneling magnetoresistive sensor, a giant magnetoresistive sensor, or a Hall-effect sensor.

### Claim 3 (depends on 1)

The anomaly detector for energy systems according to claim 1, wherein each electric field sensor is a microelectromechanical sensor, a fiber-optic electric field sensor, or an electric field meter.

### Claim 4 (depends on 1)

The anomaly detector for energy systems according to claim 1, wherein the data clustering is performed using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) using a selected minimum number of points to form a cluster and a parameter indicating the distance between different clusters based on a dataset feature from magnetic and electric-field measurements.

### Claim 5 (depends on 1)

The anomaly detector for energy systems according to claim 1, further comprising a wireless transmitter to connect to a host device.

### Claim 6 (depends on 1)

The anomaly detector for energy systems according to claim 1, wherein the anomaly indicator provides the identified anomaly to a user of the anomaly detector to enable the user to take corrective action in the energy system based on the identified anomaly.

### Claim 7

A method for anomaly detection in energy systems, the method comprising:
disposing a non-contact magnetic field sensor and a non-contact electric field sensor adjacent to an energy system;
performing non-contact sensing of a magnetic field of the energy system using the non-contact magnetic field sensor to produce a magnetic field signal indicating the operating current in the energy system;
performing non-contact sensing of an electric field of the energy system using the non-contact electric field sensor to produce an electric field signal indicating the operating voltage in the energy system;
filtering the magnetic field signal and the electric field signal to remove electromagnetic fields in the background that are not generated from the energy system;
extracting features from the magnetic field signal and the electric field signal to characterize parameters of each signal;
performing data clustering of extracted features; and
identifying an anomaly from one or more data points that do not fall within a cluster indicating normal conditions.

### Claim 8 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the data clustering of extracted features is performed using a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm that uses a selected minimum number of points to form a cluster and a parameter indicating a distance between different clusters based on a dataset feature.

### Claim 9 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the energy system is a transmission system or a residential electrical cable.

### Claim 10 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the magnetic field sensor is an anisotropic magnetoresistive sensor, a tunneling magnetoresistive sensor, a giant magnetoresistive sensor, or a Hall-effect sensor.

### Claim 11 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the electric field sensor is a microelectromechanical sensor, a fiber-optic electric field sensor, or an electric field meter.

### Claim 12 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the feature extraction includes analysis of a Fourier frequency spectrum of the electric field signal and/or the magnetic field signal.

### Claim 13 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the feature extraction includes analyzing a maximum, minimum, and root-mean-square value of electric-and magnetic-field measurement of electromagnetic sensors in a time cycle, and a frequency spectrum in fundamental and harmonic frequency.

### Claim 14 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the non-contact sensing of the magnetic field of the energy system employs an array of non-contact magnetic field sensors.

### Claim 15 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, wherein the non-contact sensing of the electric field of the energy system employs an array of non-contact electric field sensors.

### Claim 16 (depends on 7)

The method for anomaly detection in energy systems according to claim 7, further comprising:
taking corrective action in the energy system based on the identified anomaly.

### Claim 17 (depends on 16)

The method for anomaly detection in energy systems according to claim 16, wherein the magnetic field sensor is an anisotropic magnetoresistive sensor, a tunneling magnetoresistive sensor, a giant magnetoresistive sensor, or a Hall-effect sensor, and
wherein the electric field sensor is a microelectromechanical sensor, a fiber-optic electric field sensor, or an electric field meter.

### Claim 18 (depends on 6)

The anomaly detector for energy systems according to claim 6, wherein each magnetic field sensor is an anisotropic magnetoresistive sensor, a tunneling magnetoresistive sensor, a giant magnetoresistive sensor, or a Hall-effect sensor, and
wherein each electric field sensor is a microelectromechanical sensor, a fiber-optic electric field sensor, or an electric field meter.