Granted Patent B2

US 12,341,536 · App. 18/255,291 · Granted Jun 24, 2025

Inventors: Daniel Zucchetto (Dublin, IE); Johannes Van De Belt (Dublin, IE); Niall Cahill (Dublin, IE); Keith Nolan (Westmeath, IE)

Assignee: EATON INTELLIGENT POWER LIMITED

H03M7/3068H03M7/3059

Abstract

Claims (37)

1. A data transmission and compression method arranged to compress and transmit data between an edge device and a remote server, the method comprising:

collecting data at the edge device, wherein the data comprises a plurality of data signatures, wherein the collected data signatures comprise measurement data from a plurality of N correlated sensors at the edge device, and wherein the data signatures are periodic signals collected as a time series;

generating an N-dimensional data matrix at the edge device, comprising incrementing by one a position in the data matrix every time instant collected data values from the N correlated sensors correspond to that position over a predefined time duration, and scaling the values of the data matrix between zero and one depending on a maximum value recorded in the data matrix;

transforming the data matrix, wherein transforming the data matrix comprises using a wavelet transform and wherein the step of transforming the data comprises sparsifying the data matrix to obtain a sparsified data matrix;

compressing data of the sparsified data matrix to obtain encoded compressed data, wherein compressing the data comprises utilizing an autoencoder;

transmitting the encoded compressed data to the remote server via a communication channel;

at the remote server, decompressing the encoded compressed data utilizing an autoencoder;

reconstructing the data signatures using an inverse wavelet transform to obtain reconstructed signal signatures; and,

storing the reconstructed signal signatures in a datastore on the remote server.

2. The method of claim 1 , wherein the collected data signatures comprise voltage and current signals.

3. The method of claim 1 , wherein the data collected is measured raw data.

4. The method of claim 1 , wherein the time-series compromises difference values (Δx) for succeeding measurements, such that x 2 =x 1 +Δx 1 , x 3 =x 2 +Δx 2 , x 4 , =x 3 +Δx 3 , wherein Δx can be a positive or negative integer.

5. The method of claim 1 , wherein the edge device comprises an embedded device.

6. The method of claim 1 , wherein the transmission is over a network protocol.

7. The method of claim 1 , wherein the data is collected from a PowerGenome architecture.

8. The method of claim 1 , wherein the autoencoder on the edge device comprises a neural network with an input layer.

9. The method of claim 1 , wherein the autoencoder on the remote server comprises a pre-trained neural network.

10. A data transmission and compression system arranged to compress and transmit data via a communication channel between an edge device comprising a plurality of N correlated sensors and a remote server, comprising:

an edge device, wherein the edge device is configured to:

collect a plurality of data signatures, wherein the collected data signatures comprise measurement data from a plurality of N correlated sensors at the edge device, and wherein the data signatures are periodic signals collected as a time series,

generate an N-dimensional data matrix, increment by one a position in the data matrix every time instant collected data values from the N correlated sensors correspond to that position over a predefined time duration, and scale the values of the data matrix between zero and one depending on a maximum value recorded in the data matrix,

transform data of the data matrix using a wavelet transform and sparsify the data matrix to obtain a sparsified data matrix, and

compress data of the sparsified data matrix utilizing an autoencoder to obtain encoded compressed data,

wherein the edge device includes a transmitter that transmits the encoded compressed data to the remote server via the communication channel; and

a remote server comprising a receiver, wherein the remote server is configured to:

decompress the data using an autoencoder to obtain decompressed data, and

reconstruct the data signatures using inverse wavelet transform to obtain reconstructed data signatures,

wherein the remote server includes a datastore and reconstructed data signatures are stored in the data store.

11. A data transmission and compression method arranged to compress and transmit data between an edge device and a remote server, the method comprising:

collecting data at the edge device, wherein the data includes a plurality of data signatures, wherein the collected data signatures comprise measurement data from a plurality of N correlated sensors at the edge device, and wherein the data signatures are periodic signals collected as a time series;

generating an N-dimensional data matrix at the edge device, wherein the data matrix has a number of rows corresponding to a number of time instants at which data is collected over a predefined time duration, and wherein generating the data matrix includes scaling the values of the data matrix between zero and one depending on a maximum value recorded in the data matrix;

transforming the data matrix, wherein transforming the data matrix comprises using a wavelet transform and wherein the step of transforming the data includes sparsifying the data matrix to obtain a sparsified data matrix;

compressing data of the sparsified data matrix to obtain encoded compressed data, wherein compressing the data comprises utilizing an autoencoder;

transmitting the encoded compressed data to the remote server via a communication channel;

the method further comprising, at the remote server, decompressing the encoded compressed data utilizing an autoencoder;

reconstructing the data signatures using an inverse wavelet transform to obtain reconstructed signal signatures; and,

storing the reconstructed signal signatures in a datastore on the remote server.

Assignments (1)

Continuity (1)

References Cited (26)