System and method for electric load recognition from centrally monitored power signal and its application to home energy management

• US 8,560,134 B1
• Filed: 09/10/2010
• Issued: 10/15/2013
• Est. Priority Date: 09/10/2010
• Status: Expired due to Fees

• Current Assignee: Kwangduk Douglas Lee
• Original Assignee: Kwangduk Douglas Lee
• Inventors: Lee, Kwangduk Douglas
• Primary Examiner(s): ORTIZ RODRIGUEZ, CARLOS R
• Application Number:
  US12/879,983
• Time in Patent Office: 1,131 Days
• Field of Search: 700 28- 32, 700 47- 50, 700/286, 700/291, 700/297
• US Class Current: 700/291
• CPC Class Codes: G06Q 50/06 Energy or water supply
• CPC Class Codes: H02J 13/00002 characterised by monitoring
• CPC Class Codes: H02J 13/00026 involving a local wireless network, e.g. Wi-Fi, ZigBee or Bluetooth
• CPC Class Codes: H02J 13/00028 involving the use of Internet protocols
• CPC Class Codes: H02J 2310/70 Load identification
• CPC Class Codes: Y02B 90/20 Smart grids as enabling technology in buildings sector smart grids supporting the management or operation of end-user stationary applications in general, or like technologies with no associated climate change mitigation effect Y04S20/00
• CPC Class Codes: Y02E 40/70 Smart grids as climate change mitigation technology in the energy generation sector
• CPC Class Codes: Y02E 60/00 Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
• CPC Class Codes: Y04S 10/30 State monitoring, e.g. fault, temperature monitoring, insulator monitoring, corona discharge
• CPC Class Codes: Y04S 10/50 Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
• CPC Class Codes: Y04S 40/126 using wireless data transmission
• CPC Class Codes: Y04S 40/128 involving the use of Internet protocol

Abstract

A method and apparatus are provided for a home energy management platform. The platform includes using a whole house power sensor or subset thereof. Data from the power sensor are analyzed using advanced statistical and machine learning techniques for extracting detailed usage information and generating specific energy saving measures, among other relevant information. In an embodiment, a gateway console is provided that has various communication capabilities. The gateway console may communicate with and control HAN devices. The gateway console may collect data from the power sensor as well as HAN devices and upload such collected data to servers for the analysis processing. Certain amounts of data processing and analysis may be performed at a server or at the local level, such as at the power sensor, gateway, or other HAN device, as well. The platform may include a user interface, such as web, mobile, email, mail, phone call, etc.

Claims

1. A computer-implemented method for load recognition from a monitored signal and application to energy management service, comprising the steps of:

extracting, by a feature extraction algorithm processor, features from a set of power data using a feature extraction algorithm to obtain a plurality of events, wherein the set of power data reflects power consumed by one or more loads;

initializing, by an initializing processor, a load library repository with initial load instance data, the initial load instance data comprising;

initial estimates of state probabilities and state transition probabilities that represent a plurality of loads;

performing a load recognition algorithm, by a load recognition algorithm processor, said load recognition algorithm comprising the steps of;

for an event x from said plurality of events;

assigning, by an assigning processor, said event x and particular subsequent events to a particular load,

wherein the assigning is based on said particular load giving a best posterior probability from a best posterior probability algorithm performed on said plurality of events, and

wherein said assigning is performed after performing a search and classification algorithm on said plurality of events; and

adding, by an adding processor, load instance data associated with said particular load to said load library repository and continuing the adding load instance data for a next available event from said plurality of events; and

updating based on said adding of load instance data, by an updating processor, said load library repository with new state probabilities and state transition probabilities,

repeating said load recognition algorithm using the updated load library repository until a particular criterion is met; and

using, by an energy management service processor, said added load instance data in providing energy consuming feedback and usage analysis to be used for energy management service;

wherein said updated load library repository is used in a continuous mode on a continuous stream of power data from the plurality of loads and wherein when said added load instance data is added, combining said added load instance data with said initial load instance data to compute said new state probabilities and state transition probabilities from said combination;

said continuous mode further comprising the steps of;

extracting in real time, by said feature extraction algorithm, features from a window of said stream of power data for eligible events of said plurality of events in said window, performing a roughly estimated assignment to the particular load based on a state and state transition probability of each eligible event only;

if there are not enough subsequent events for each roughly assigned event from said window and past windows of power data, then return to said extracting in real time step, otherwise;

performing load recognition, by said load recognition algorithm, to a roughly assigned event with enough subsequent events based on a highest posterior probability search and classification and creating load instance recognition data;

marking, by a marking processor, said roughly assigned event with said enough subsequent events as used;

adding, by said adding processor, said load instance recognition data to said load library repository;

updating based on the added load instance recognition, by said updating processor, state and state transition probability estimates for said particular load data; and

with the arrival of a new window of power data, repeating said steps beginning from said step of extracting in real time features from said window of stream of power data;

wherein said plurality of loads comprise a plurality of appliances.

2. The computer implemented method of claim 1, wherein said particular load is any of electric, gas, and water.

3. The computer implemented method of claim 1, wherein said particular load is electric and said stream of power data is from a power sensor and is received at any of said power sensor itself, a gateway console that is in communication with said power sensor, and a server when said gateway console is in communication with said server and sends said stream of power data to said server.

4. The computer implemented method of claim 3, wherein said feature extraction algorithm is performed at any of said power sensor, said gateway console, and server.

5. The computer implemented method of claim 3, wherein said search and classification algorithm is performed at any of said power sensor, said gateway console, and said server.

6. The computer implemented method of claim 3,

wherein said energy management service processor or parts of said energy management service processor are embedded in any of said power sensor, said gateway console, and said server.

7. The computer implemented method of claim 2, wherein said energy management service processor is configured to target particular appliances for energy efficiency, demand response service, and time of use related energy saving tips by performing any of:

monitoring and controlling various networked appliances,

integrating renewable generation and storage of energy,

providing security,

providing site automation service,

integrating site entertainment function, and

controlling how utilities are consumed, generated, and stored.

8. The computer implemented method of claim 1, wherein said energy consuming feedback is made available to one or more consumers via any of:

a web interface, a mobile device interface, e-mail, mail, and a telephone call.

9. The computer implemented method of claim 1, wherein said added load instance data comprises any of profile information about a particular site, time, duration, and energy used.

10. The computer implemented method of claim 1, wherein said feature extraction comprises detecting and computing changes of mean and variance at each segment boundary of said signal wherein said computed changes of mean and variance are used in maximum likelihood estimates of unknown load events.

11. The computer implemented method of claim 3, wherein load states are grouped based on how one or more electric components of a particular appliance are physically used to accomplish a mission and a resulting power consumption behavior change of the particular appliance.

12. The computer implemented method of claim 1, wherein said search and classification algorithm prunes impossible sequences from a search stack based on a state probability and state transition probability and uses a time limit for the particular load.

13. The computer implemented method of claim 1, wherein said best posterior probability is computed for a load i, given x, as the product of likelihood and prior probability, where p(x) is a normalization factor:

14. The computer implemented method of claim 1, further comprising the steps of:

computing an energy fitness number, f, defined by;

15. A system for load recognition from a monitored signal and application to energy management service, comprising:

a feature extraction algorithm processor configured for extracting features from a set of power data using a feature extraction algorithm to obtain a plurality of events, wherein the set of power data reflects power consumed by one or more loads;

an initializing processor configured for initializing a load library repository with initial load instance data, the initial load instance data comprising;

initial estimates of state probabilities and state transition probabilities that represent a plurality of loads;

a load recognition algorithm processor configured for performing a load recognition algorithm, comprising the steps of;

for an event x from said plurality of events;

assigning, by an assigning processor, said event x and particular subsequent events to a particular load,

wherein the assigning is based on said particular load giving a best posterior probability from a best posterior probability algorithm performed on said plurality of events, and

wherein said assigning is performed after performing a search and classification algorithm on said plurality of events; and

adding, by an adding processor, load instance data associated with said particular load to said load library repository and continuing the adding load instance data for a next available event from said plurality of events; and

updating based on said adding of load instance data, by an updating processor, said load library repository with new state probabilities and state transition probabilities,

repeating said load recognition algorithm using the updated load library repository until a particular criterion is met; and

using, by an energy management service processor, said added load instance data in providing energy consuming feedback and usage analysis to be used for energy management service;

wherein said updated load library repository is used in a continuous mode on a continuous stream of power data from the plurality of loads and wherein when said added load instance data is added, combining said added load instance data with said initial load instance data to compute said new state probabilities and state transition probabilities from said combination;

said continuous mode further comprising the steps of;

extracting in real time, by said feature extraction algorithm, features from a window of said stream of power data for eligible events of said plurality of events in said window, performing a roughly estimated assignment to the particular load based on a state and state transition probability of each eligible event only;

if there are not enough subsequent events for each roughly assigned event from said window and past windows of power data, then return to said extracting in real time step, otherwise;

performing load recognition, by said load recognition algorithm, to a roughly assigned event with enough subsequent events based on a highest posterior probability search and classification and creating load instance recognition data;

marking, by a marking processor, said roughly assigned event with said enough subsequent events as used;

adding, by said adding processor, said load instance recognition data to said load library repository;

updating based on the added load instance recognition, by said updating processor, state and state transition probability estimates for said particular load data; and

with the arrival of a new window of power data, repeating said steps beginning from said step of extracting in real time features from said window of stream of power data;

wherein said plurality of loads comprise a plurality of appliances.

16. A non-transitory machine-readable storage medium storing one or more sequences of instructions for load recognition from a monitored signal and application to energy management service, which instructions, when executed by one or more processors, cause the one or more processors to carry out the steps of:

extracting, by a feature extraction algorithm processor, features from a set of power data using a feature extraction algorithm to obtain a plurality of events, wherein the set of power data reflects power consumed by one or more loads;

initializing, by an initializing processor, a load library repository with initial load instance data, the initial load instance data comprising;

initial estimates of state probabilities and state transition probabilities that represent a plurality of loads;

performing a load recognition algorithm, by a load recognition algorithm processor, said load recognition algorithm comprising the steps of;

for an event x from said plurality of events;

assigning, by an assigning processor, said event x and particular subsequent events to a particular load,

wherein the assigning is based on said particular load giving a best posterior probability from a best posterior probability algorithm performed on said plurality of events, and

wherein said assigning is performed after performing a search and classification algorithm on said plurality of events; and

adding, by an adding processor, load instance data associated with said particular load to said load library repository and continuing the adding load instance data for a next available event from said plurality of events; and

updating based on said adding of load instance data, by an updating processor, said load library repository with new state probabilities and state transition probabilities,

repeating said load recognition algorithm using the updated load library repository until a particular criterion is met; and

using, by an energy management service processor, said added load instance data in providing energy consuming feedback and usage analysis to be used for energy management service;

wherein said updated load library repository is used in a continuous mode on a continuous stream of power data from the plurality of loads and wherein when said added load instance data is added, combining said added load instance data with said initial load instance data to compute said new state probabilities and state transition probabilities from said combination;

said continuous mode further comprising the steps of;

extracting in real time, by said feature extraction algorithm, features from a window of said stream of power data for eligible events of said plurality of events in said window, performing a roughly estimated assignment to the particular load based on a state and state transition probability of each eligible event only;

if there are not enough subsequent events for each roughly assigned event from said window and past windows of power data, then return to said extracting in real time step, otherwise;

performing load recognition, by said load recognition algorithm, to a roughly assigned event with enough subsequent events based on a highest posterior probability search and classification and creating load instance recognition data;

marking, by a marking processor, said roughly assigned event with said enough subsequent events as used;

adding, by said adding processor, said load instance recognition data to said load library repository;

updating based on the added load instance recognition, by said updating processor, state and state transition probability estimates for said particular load data; and

with the arrival of a new window of power data, repeating said steps beginning from said step of extracting in real time features from said window of stream of power data;

wherein said plurality of loads comprise a plurality of appliances.

17. A computer implemented method for searching and for classifying events for load recognition, comprising the steps of:

iterating over available events;

for a given event x of said available events, searching, by a searching processor, possible state transition sequences for a load i, wherein subsequent events after said given event x are considered in finding the state transition sequences for said load, wherein said given event x and said subsequent events are referred to collectively as vector x, and;

pruning, by a pruning processor, impossible state transition sequences from a particular search stack, based on state or based on state transition probability for a particular time limit and until possible state sequences are found;

computing, by a computing processor, a likelihood xi, for candidates and choosing a state sequence with a maximum xi;

computing a posterior probability of said load i, for said vector x, wherein the posterior probability of load i is a product of likelihood and prior probability and p(x) is a normalization factor according to;