Power consumption analyzing server and power consumption analyzing method thereof

Patent metadata

Field	Value
Publication number	US-10120004-B1
Application number	US-201715814714-A
Country	US
Kind code	B1
Filing date	Nov 16, 2017
Priority date	Oct 20, 2017
Publication date	Nov 6, 2018
Grant date	Nov 6, 2018

Abstract

A power consumption analyzing server and a power consumption analyzing method thereof are provided. According to the user data, the power consumption analyzing server clusters users into a plurality of groups. For each group, the power consumption analyzing server generates M+1 number of encoded outputs by inputting total power consumption time sequence data corresponding to a reference user in the group into an autoencoder. For each group, the power consumption analyzing server receives an actual appliance measurement data from an energy management system apparatus of the reference user, and labels M number of encoded outputs of the M+1 number of encoded outputs to map them to M categories of appliances. Finally, for each group, the power consumption analyzing server establishes a non-intrusive load monitoring system model of the group.

First claim

What is claimed is:

1. A power consumption analyzing server, comprising:
a network interface;
a storage, being configured to store a user database that records a user datum of each of
a plurality of users;

a processor electrically connected with the network interface and the storage, being configured to execute the following steps:
(a) clustering the users into a plurality of groups according to the user data;
(b) for each of the groups, generating M+1 encoded outputs by inputting a total power consumption time sequence datum of a reference user corresponding to the group among the user data into an autoencoder, where M is a positive integer;
(c) for each of the groups, receiving an actual appliance measurement datum from an energy management system apparatus of the reference user in the group by connecting to the energy management system apparatus via the network interface, the actual appliance measurement datum being associated with M categories of appliances;
(d) for each of the groups, labeling M encoded outputs among the M+1 encoded outputs according to the actual appliance measurement datum to map the M encoded outputs to the M categories of appliances; and
(e) for each of the groups, establishing a non-intrusive load monitoring system (NILM) model of the group according to the M+1 encoded outputs and the actual appliance measurement datum after the M encoded outputs are labeled.

2. The power consumption analyzing server of claim 1 , wherein each of the user data further comprises at least one of: statistical power consumption, a load curve, a building category, a population in a house, a region category, and a climate category.

3. The power consumption analyzing server of claim 1 , wherein the total power consumption time sequence data are measured at a sampling rate of lower than or equal to one sample per minute.

4. The power consumption analyzing server of claim 1 , wherein for a served user among the users, the processor further generates a power consumption suggestion datum according to the non-intrusive load monitoring system model of the group.

5. The power consumption analyzing server of claim 1 , wherein for each of the groups, the processor further corrects the non-intrusive load monitoring system model of the group according to the actual appliance measurement data of the reference user.

6. The power consumption analyzing server of claim 1 , wherein the processor performs one-dimensional, two-dimensional or multi-dimensional clustering according to at least one parameter in the user data to cluster the users into the groups.

7. The power consumption analyzing server of claim 1 , wherein the processor further connects to an electricity data server via the network interface to obtain the user data of the users from the electricity data server.

8. The power consumption analyzing server of claim 1 , wherein the processor further connects to an electricity data server via the network interface to obtain a total power consumption datum of each of the users from the electricity data server, wherein the total power consumption time sequence datum of the reference user is obtained from the total power consumption datum of the reference user by setting an observation window.

9. The power consumption analyzing server of claim 8 , wherein the total power consumption datum is constituted by a plurality of sub-total power consumption time sequence data, and any two of the sub-total power consumption time sequence data that are adjacent in time sequence are partly overlapped or non-overlapped with each other.

10. The power consumption analyzing server of claim 1 , wherein the processor further connects to a plurality of smart electricity meters via the network interface.

11. A power consumption analyzing method for a power consumption analyzing server, the power consumption analyzing server comprising a processor, a network interface and a storage, the storage being configured to store a user database that records a user datum of each of a plurality of users, the power consumption analyzing method being executed by the processor and comprising the following steps:
(a) clustering the users into a plurality of groups according to the user data;
(b) for each of the groups, generating M+1 encoded outputs by inputting a total power consumption time sequence datum of a reference user corresponding to the group among the user data into an autoencoder, where M is a positive integer;
(c) for each of the groups, receiving an actual appliance measurement datum from an energy management system apparatus of the reference user in the group by connecting to the energy management system apparatus via the network interface, the actual appliance measurement datum being associated with M categories of appliances;
(d) for each of the groups, labeling M encoded outputs among the M+1 encoded outputs according to the actual appliance measurement datum to map the M encoded outputs to the M categories of appliances; and
(e) for each of the groups, establishing a non-intrusive load monitoring system (NILM) model of the group according to the M+1 encoded outputs and the actual appliance measurement datum after the M encoded outputs are labeled.

12. The power consumption analyzing method of claim 11 , wherein each of the user data further comprises at least one of: statistical power consumption, a load curve, a building category, a population in a house, a region category, and a climate category.

13. The power consumption analyzing method of claim 11 , wherein the total power consumption time sequence data are measured at a sampling rate of lower than or equal to one sample per minute.

14. The power consumption analyzing method of claim 11 , further comprising the following step:
for a served user among the users, generating a power consumption suggestion datum according to the non-intrusive load monitoring system model of the group corresponding to the served user.

15. The power consumption analyzing method of claim 11 , further comprising the following step:
for each of the groups, correcting the non-intrusive load monitoring system model of the group according to the actual appliance measurement data of the reference user.

16. The power consumption analyzing method of claim 11 , wherein the step (a) is to perform one-dimensional, two-dimensional or multi-dimensional clustering according to at least one parameter in the user data to cluster the users into the groups.

17. The power consumption analyzing method of claim 11 , further comprising the following step:
connecting to an electricity data server via the network interface to obtain the user data of the users from the electricity data server.

18. The power consumption analyzing method of claim 11 , wherein the power consumption analyzing server connects to an electricity data server to obtain a total power consumption datum of each of the users from the electricity data server, wherein the total power consumption time sequence datum of the reference user is obtained from the total power consumption datum of the reference user by setting an observation window.

19. The power consumption analyzing method of claim 18 , wherein the total power consumption datum is constituted by a plurality of sub-total power consumption time sequence data, and any two of the sub-total power consumption time sequence data that are adjacent in time sequence are partly overlapped or non-overlapped with each other.

20. The power consumption analyzing method of claim 11 , wherein the power consumpt

Assignees

• Inst Information Ind

Inventors

• Tsai Chia-Wei
• Hung Yung-Chieh
• Chiang Kuei-Chun

Classifications

• H02J13/1337

  involving the use of Internet protocols · CPC title

• H02J13/12

  Monitoring network conditions, e.g. electrical magnitudes or operational status · CPC title

• G06F15/76

  Architectures of general purpose stored program computers (with program plugboard G06F15/08; multicomputers G06F15/16) · CPC title

• H02J3/00

  Circuit arrangements for AC mains or AC distribution networks · CPC title

• H05B47/155

  Coordinated control of two or more light sources · CPC title