System and method for monitoring resources in a water utility network

• US 9,568,392 B2
• Filed: 03/04/2011
• Issued: 02/14/2017
• Est. Priority Date: 03/04/2010
• Status: Active Grant

• Current Assignee: TaKaDu Ltd.
• Original Assignee: TaKaDu Ltd.
• Inventors: Peleg, Amir, Armon, Amitai, Barkay, Uri, Scolnicov, Haggai, Gutner, Shai
• Primary Examiner(s): WONG, ALBERT KANG
• Application Number:
  US13/040,435
• Publication Number:
  US 20110215945A1
• Time in Patent Office: 2,174 Days
• Field of Search: 702/50, 702/100, 702/51, 34087002-87003, 340/605, 73/40, 73/405, 137/1
• US Class Current: 1/1
• CPC Class Codes: G01M 3/2807 for pipes G01M3/2892, G01M3/30 take precedence
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 50/06 Energy or water supply
• 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: Y04S 20/30 Smart metering, e.g. specially adapted for remote reading
• CPC Class Codes: Y10T 137/0318 Processes

Abstract

A computerized method for monitoring a water utility network, the water utility network comprising a network of pipes for delivering water to consumers and a plurality of meters positioned within the pipes across the water distribution network. The method includes receiving meter data representing parameters measured by the meters, such as flow, pressure, chlorine level, pH and turbidity of the water being distributed through the pipes. The method also includes receiving secondary data from sources external to the meters and representing conditions affecting consumption of water in a region serviced by the water utility network such as weather and holidays. The meter and secondary data is analyzed using statistical techniques to identify water network events including leakage events and other events regarding quantity and quality of water flowing through the pipes and operation of the water network. The events are reported to users via a user interface.

Claims

1. A computerized method for analyzing resource distribution throughout a network, the method comprising:

receiving meter data, the meter data representing a plurality of parameters measured by a plurality of meters positioned within a resource distribution network;

receiving secondary data from one or more sources external to the meters, the secondary data representing one or more conditions affecting a condition of network assets in a region serviced by the resource distribution network;

statistically predicting meter data for at least one meter based on secondary data;

comparing the received meter data with the likely distribution of predicted values to detect one or more anomalies if the received meter data deviates from the likely distribution of predicted values;

identifying a resource distribution network event on the basis of the detected anomalies wherein identifying the resource distribution network event comprises identifying a faulty meter event by;

receiving a first data set representative of a first meter to be analyzed,

receiving a reference data set representative of a physically separate second meter,

correlating the first data set with the reference data set,

determining a time shift based on the correlation of the first data set with the reference data set, and

issuing a faulty meter event based on a statistical significance of the time shift; and

reporting the resource distribution network event.

2. The method of claim 1, comprising:

selecting at least one parameter from the plurality of parameters; and

predicting, based on linear regression and a root mean square error metric, the likely distribution of predicted values using a function of the selected at least one parameter that approximates the predicted values.

3. The method of claim 1, wherein statistically predicting meter data comprises identifying historical data associated with a second meter comprised of:

selecting a data set generated from the second meter among available data sets generated from other meters; and

minimizing a prediction error by analyzing all possible combinations of available data sets.

4. The method of claim 3, wherein selecting a data set among available data sets comprises:

finding a best fit parameter of the selected data set;

finding an approximation error achieved of the selected data set; and

selecting at least one data set among the available data sets with the lowest approximation error.

5. The method of claim 1, wherein receiving meter data includes receiving data selected from the set consisting of:

an individual meter, a collection of meters, and a collection of meters measuring a same type of data.

6. The method of claim 1, further comprising associating data with each network event, the associated data selected from the set consisting of:

a start time of the event, an end time of the event, a magnitude of the event, an accumulated magnitude of the event, and a proportion of the event.

7. The method of claim 1, further comprising detecting an anomaly if an area under a curve of a difference between the received meter value and a predicted meter value exceeds a predefined threshold.

8. The method of claim 1, further comprising associating one or more resource distribution network events with each other based on an event type, a start time of the one or more resource distribution network events, or location of the one or more resource distribution network events.

9. The method of claim 8, further comprising using the association between one or more resource distribution network events to classify a start of an event, an end of an event, or a relationship between events.

10. The method of claim 1, further comprising:

associating the one or more anomalies with a statistical confidence level;

adjusting the statistical confidence level based on the secondary data; and

classifying the anomaly as a resource distribution network event based on the statistical confidence level exceeding a threshold.

11. The method of claim 1, further comprising identifying an unexpected consumption event, wherein identifying the unexpected consumption event comprises:

detecting an increased flow reading of a meter;

calculating a statistical significance of at least one of the increased flow reading of the given meter, a consumption pattern, a reoccurrence of the increased flow reading, and a magnitude of each reoccurrence of the increased flow reading; and

comparing the calculated statistical significance to the secondary data, the secondary data used to verify or refute the unexpected consumption event.

12. The method of claim 1, further comprising identifying a water quality event, wherein identifying the water quality event comprises:

identifying water quality data present within the meter data, wherein the water quality includes at least one of measured chlorine, turbidity, and pH;

detecting a statistically significant change in the water quality data; and

comparing the detected statistically significant change to the secondary data, the secondary data used to verify or refute the water quality event.

13. The method of claim 1 further comprising formatting the received meter data for further analysis.

14. The method of claim 13 wherein formatting the received meter data comprises removing data not reflecting the actual performance of the resource distribution network.

15. The method of claim 13 wherein formatting the received meter data comprises removing selected resource distribution network phenomena or phenomena related to intentional operation of the network.

16. The method of claim 13 wherein formatting the received meter data comprises smoothing data collected from the resource distribution network.

17. The method of claim 16 wherein smoothing data collected from the resource distribution network comprises at least one of:

Locally Weighted Scatterplot Smoothing and heuristic cleaning.

18. The method of claim 1 further comprising filtering the received meter data to remove meter transmission noise.

30. The method of claim 1, wherein identifying a faulty meter event further comprises detecting a fixed value transmission over time.

31. The method of claim 1, wherein identifying a faulty meter event further comprises detecting a statistically unlikely magnitude of time-dependent variability.

32. The method of claim 1, wherein identifying a faulty meter event further comprises:

calculating a frequency of data currently transmitted by a first meter;

calculating an amount of data historically transmitted by the first meter or a second meter; and

determining a statistically significant deviation of the amount of data currently transmitted and the amount of data historically transmitted.

33. The method of claim 1, wherein identifying a faulty meter event further comprises:

determining a supported value for a meter under analysis, the supported value based on a second meter in the network; and

comparing the supported value to an actual value produced by the meter under analysis.

19. A non-transitory computer-readable medium comprising machine-readable instructions that when executed by a computer causes the computer to perform a method for analyzing resource distribution throughout a network, the computer-readable medium comprising:

computer program code for receiving meter data, the meter data representing a plurality of parameters measured by a plurality of meters positioned within a resource distribution network;

computer program code for receiving secondary data from one or more sources external to the meters, the secondary data representing one or more conditions affecting a condition of network assets in a region serviced by the resource distribution network;

computer program code for statistically predicting meter data for at least one meter based on secondary data;

computer program code for comparing the received meter data with the predicted meter data to detect one or more anomalies if the received meter data deviates from the predicted meter data;

computer program code for identifying a resource distribution network event on the basis of the detected anomalies wherein the computer program code for identifying the resource distribution network event comprises computer program code for identifying a faulty meter event by;

receiving a first data set representative of a first meter to be analyzed,

receiving a reference data set representative of a physically separate second meter,

correlating the first data set with the reference data set,

determining a time shift based on the correlation of the first data set with the reference data set, and

issuing a faulty meter event based on a statistical significance of the time shift; and

computer program code for reporting the resource distribution network event.

20. The computer-readable medium of claim 19, further comprising:

computer program code for selecting at least one parameter from the plurality of parameters; and

predicting, based on linear regression and a root mean square error metric, the likely distribution of predicted values using a function of the selected at least one parameter that approximates the predicted values.

21. The computer-readable medium of claim 19, wherein statistically predicting meter data comprises:

computer program code for selecting a data set generated from the second meter among available data sets generated from other meters; and

computer program code for minimizing a prediction error by analyzing all possible combinations of available data sets.

22. The computer-readable medium of claim 19, wherein selecting the data set comprises:

computer program code for finding a best fit parameter of the selected data set;

computer program code for finding an approximation error achieved of the selected data set; and

computer program code for selecting at least one data set among the available data sets with the lowest approximation error.

23. The computer-readable medium of claim 19, wherein the computer program code for identifying a faulty meter event further comprises computer program code for detecting a fixed value transmission over time.

24. The computer-readable medium of claim 19, wherein the computer program code for identifying a faulty meter event further comprises computer program code for detecting a statistically unlikely magnitude of time-dependent variability.

25. The computer-readable medium of claim 19, wherein the computer program code for identifying a faulty meter event further comprises:

computer program code for calculating a frequency of data currently transmitted by a first meter;

computer program code for calculating an amount of data historically transmitted by the first meter or a second meter; and

computer program code for determining a statistically significant deviation of the amount of data currently transmitted and the amount of data historically transmitted.

26. The computer-readable medium of claim 19, wherein the computer program code for identifying a faulty meter event further comprises:

computer program code for determining a supported value for a meter under analysis, the supported value based on a second meter in the network; and

computer program code for comparing the supported value to an actual value produced by the meter under analysis.

27. A computerized system for analyzing resource distribution throughout a network the system comprising:

a database for storing meter data, the meter data representing a plurality of parameters measured by a plurality of meters positioned within a resource distribution network and secondary data from one or more sources external to the meters, the secondary data representing one or more conditions affecting a condition of network assets in a region serviced by the resource distribution network;

an analysis engine configured to analyze the meter data by statistically predicting meter data for at least one meter based on secondary data and comparing the stored meter data with the predicted meter data to detect one or more anomalies if the stored meter data deviates from the predicted meter data;

an event classification engine configured to identify a resource distribution network event based on the one or more anomalies by;

receiving a first data set representative of a first meter to be analyzed,

receiving a reference data set representative of a physically separate second meter,

correlating the first data set with the reference data set,

determining a time shift based on the correlation of the first data set with the reference data set, and

issuing a faulty meter event based on a statistical significance of the time shift; and

an event database for storing resource distribution network event data representing the resource distribution network event identified by the classification engine.

28. The system of claim 27, wherein the analysis engine selects at least one parameter from the plurality of parameters;

and

predicts, based on linear regression and a root mean square error metric, the likely distribution of predicted values using a function of the selected at least one parameter that approximates the predicted values.

29. The system of claim 27, wherein, in order to predict meter data, the analysis engine:

selects a data set generated from the second meter among available data sets generated from other meters; and

minimizes a prediction error by analyzing all possible combinations of available data sets.

34. A computerized method for analyzing resource distribution throughout a network, the method comprising:

receiving meter data, the meter data representing a plurality of parameters measured by a plurality of meters positioned within a resource distribution network;

statistically predicting meter data for at least one meter;

comparing the received meter data with the predicted meter data to detect one or more anomalies if the received meter data deviates from the predicted meter data;

identifying a resource distribution network event on the basis of the detected anomalies wherein identifying the resource distribution network event comprises identifying a faulty meter event by;

receiving a first data set representative of a first meter to be analyzed,

receiving a reference data set representative of a physically separate second meter,

correlating the first data set with the reference data set,

determining a time shift based on the correlation of the first data set with the reference data set, and

issuing a faulty meter event based on a statistical significance of the time shift; and

reporting the resource distribution network event.