Techniques for multi-stage analysis of measurement data with event stream processing

• US 10,212,024 B2
• Filed: 06/04/2018
• Issued: 02/19/2019
• Est. Priority Date: 06/02/2017
• Status: Active Grant

• Current Assignee: SAS Institute Incorporated
• Original Assignee: SAS Institute Incorporated
• Inventors: Pope, David Carter, Chisum, Blanden Scott, Farina, Brett, Konya, Mark Joseph
• Primary Examiner(s): Skripnikov, Alex
• Application Number:
  US15/997,371
• Publication Number:
  US 20180351786A1
• Time in Patent Office: 260 Days
• Field of Search: None
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: H04L 41/044 comprising hierarchical management structures
• CPC Class Codes: H04L 41/0654 using network fault recovery ring fault isolation or reconfiguration in loop networks without recovery actions by a network management system H04L12/437
• CPC Class Codes: H04L 41/0813 characterised by the conditions triggering a change of settings
• CPC Class Codes: H04L 45/28 using route fault recovery
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: H04L 67/34 involving the movement of software or configuration parameters network booting or remote initial program loading [RIPL] G06F9/4416
• CPC Class Codes: Y04S 40/00 Systems for electrical power generation, transmission, distribution or end-user application management characterised by the use of communication or information technologies, or communication or information technology specific aspects supporting them

Abstract

Various embodiments are generally directed to systems for multi-stage measurement data analysis (MMDA), such as for evaluation and/or validation of data received from a measurement device, for instance. Some embodiments are particularly directed to a MMDA system that utilizes event stream processing (ESP) to provide near real-time validation of measurement data, at least in part, by detecting losses in the measurement data. In many embodiments, the MMDA system may detect technical losses (e.g., due to equipment malfunction) and/or non-technical losses (e.g., due to compromised equipment). For example, the MMDA system may receive measurement data generated by an electrical meter and determine the electrical meter is malfunctioning by detecting a technical loss in the measurement data. In many embodiments, the MMDA system may utilize both direct and indirect measurement data transmitted via separate communication paths to provide near real-time validation of measurement data.

Claims

1. An apparatus comprising a processor and a storage to store instructions that, when executed by the processor, cause the processor to perform operations comprising:

identify direct measurement device data comprising a measured value generated by a measurement device (MD), the measured value received as part of a first data stream comprising a series of measured values over an interval in a time domain, the first data stream communicated via an event stream processing (ESP) interface of the measurement device;

identify indirect MD data comprising a reported value associated with generation by the MD, the reported value and a unique identifier of the MD received as part of a second data stream comprising a series of reported values over the interval in the time domain and the unique identifier of the MD, the series of reported values associated with generation by the MD in a staging area based on the unique identifier received in the second data stream, the staging area communicatively coupled with the MD via a neighborhood area network (NAN) and the staging area comprising a temporary storage to aggregate reported values for a group of MDs associated with a geographic area of the MD;

generate an expected value for the reported value with a first machine learning algorithm based on a class assigned to the MD and a portion of the series of reported values received in the second data stream prior to the reported value being received;

compare the reported value to the expected value and the reported value to the measured value to validate the reported value associated with generation by the MD in the staging area communicatively coupled to the MD via the NAN;

transform at least one of the direct MD data and the indirect MD data into the frequency domain to generate a spectral density for the MD;

compare the spectral density for the MD with a model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD, wherein the model spectral density is generated by a second machine learning algorithm with data from a set of MDs identified to characterize the class assigned to the MD;

identify a technical loss or a non-technical loss associated with the MD based on one or more of the validation of the reported value and the verification of the class associated with the MD; and

determine one or more corrective processes associated with the MD to perform based on whether technical loss or non-technical loss associated with the MD is identified.

2. The apparatus of claim 1, wherein the processor is caused to perform operations comprising decomposing the at least one of the direct MD data and the indirect MD data into a summation of periodic functions to transform the at least one of the direct MD data and the indirect MD data into the frequency domain to generate the spectral density for the MD.

3. The apparatus of claim 2, wherein the processor is caused to perform operations comprising:

correlate an equipment type with at least one of the periodic functions included in the summation; and

utilize the equipment type to generate the expected value for the reported value.

4. The apparatus of claim 2, wherein the processor is caused to perform operations to compare of the spectral density for the MD with the model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD comprising:

perform clustering with correlation analysis or principal component analysis of the spectral density for the MD and a plurality of additional spectral densities for the set of MDs identified to characterize the class assigned to the MD; and

determine whether the MD has similar frequency content similar to one or more of the set of MDs identified to characterize the class assigned to the MD based on the clustering.

5. The apparatus of claim 1, wherein the reported value is aggregated in a second staging area communicatively coupled to the staging area in the NAN by a regional area network (RAN) prior to receival as part of the second live steam, the second staging area comprising a temporary storage to aggregate reported values for a second group of MDs associated with a geographic region of the MD, the geographic region of the MD comprising the geographic area of the MD and a second geographic area of at least one other MD.

6. The apparatus of claim 1, wherein the processor is caused to perform operations comprising combine the first data stream and the second data stream into a data record stream associated with the MD, wherein the data record stream is provided to a database for storage, and wherein the first and second machine learning algorithms are trained with data stored in the database as part of a plurality of data record streams associated with a plurality of MDs.

7. The apparatus of claim 6, wherein the data from the set of MDs identified to characterize the class assigned to the MD used by the second machine learning algorithm to generate the model spectral density was stored in the database as part of a respective data record streams corresponding to each of the set of MDs identified to characterize the class assigned to the MD.

8. The apparatus of claim 6, wherein training at least one of the first and second machine learning algorithms utilizes one or more of support vector machines (SVMs), Fourier transforms, logistic regression, deep neural networks (DNNs), principal component analysis (PCA), clustering in the frequency domain.

9. The apparatus of claim 1, wherein the processor is caused to perform operations comprising identify a non-technical loss associated with the MD based on verification of the reported value indicating the reported value exceeds the measured value.

10. The apparatus of claim 1, wherein the processor is caused to perform operations comprising assign the class to the MD based on one or more properties of a client account that corresponds to the MD.

11. A computer-implemented method, comprising:

identifying direct measurement device data comprising a measured value generated by a measurement device (MD), the measured value received as part of a first data stream comprising a series of measured values over an interval in a time domain, the first data stream communicated via an event stream processing (ESP) interface of the measurement device;

identifying indirect MD data comprising a reported value associated with generation by the MD, the reported value and a unique identifier of the MD received as part of a second data stream comprising a series of reported values over the interval in the time domain and the unique identifier of the MD, the series of reported values associated with generation by the MD in a staging area based on the unique identifier received in the second data stream, the staging area communicatively coupled with the MD via a neighborhood area network (NAN) and the staging area comprising a temporary storage to aggregate reported values for a group of MDs associated with a geographic area of the MD;

generating an expected value for the reported value with a first machine learning algorithm based on a class assigned to the MD and a portion of the series of reported values received in the second data stream prior to the reported value being received;

comparing the reported value to the expected value and the reported value to the measured value to validate the reported value associated with generation by the MD in the staging area communicatively coupled to the MD via the NAN;

transforming at least one of the direct MD data and the indirect MD data into the frequency domain to generate a spectral density for the MD;

comparing the spectral density for the MD with a model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD, wherein the model spectral density is generated by a second machine learning algorithm with data from a set of MDs identified to characterize the class assigned to the MD;

identifying a technical loss or a non-technical loss associated with the MD based on one or more of the validation of the reported value and the verification of the class associated with the MD; and

determining one or more corrective processes associated with the MD to perform based on whether technical loss or non-technical loss associated with the MD is identified.

12. The computer-implemented method of claim 11, comprising decomposing the at least one of the direct MD data and the indirect MD data into a summation of periodic functions to transform the at least one of the direct MD data and the indirect MD data into the frequency domain to generate the spectral density for the MD.

13. The computer-implemented method of claim 12, comprising:

correlating an equipment type with at least one of the periodic functions included in the summation; and

utilizing the equipment type to generate the expected value for the reported value.

14. The computer-implemented method of claim 12, wherein comparing of the spectral density for the MD with the model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD comprises:

performing clustering with correlation analysis or principal component analysis of the spectral density for the MD and a plurality of additional spectral densities for the set of MDs identified to characterize the class assigned to the MD; and

determining whether the MD has similar frequency content similar to one or more of the set of MDs identified to characterize the class assigned to the MD based on the clustering.

15. The computer-implemented method of claim 11, wherein the reported value is aggregated in a second staging area communicatively coupled to the staging area in the NAN by a regional area network (RAN) prior to receival as part of the second live steam, the second staging area comprising a temporary storage to aggregate reported values for a second group of MDs associated with a geographic region of the MD, the geographic region of the MD comprising the geographic area of the MD and a second geographic area of at least one other MD.

16. The computer-implemented method of claim 11, comprising combining the first data stream and the second data stream into a data record stream associated with the MD, wherein the data record stream is provided to a database for storage, and wherein the first and second machine learning algorithms are trained with data stored in the database as part of a plurality of data record streams associated with a plurality of MDs.

17. The computer-implemented method of claim 16, wherein the data from the set of MDs identified to characterize the class assigned to the MD used by the second machine learning algorithm to generate the model spectral density was stored in the database as part of a respective data record streams corresponding to each of the set of MDs identified to characterize the class assigned to the MD.

18. The computer-implemented method of claim 16, wherein training at least one of the first and second machine learning algorithms utilizes one or more of support vector machines (SVMs), Fourier transforms, logistic regression, deep neural networks (DNNs), principal component analysis (PCA), clustering in the frequency domain.

19. The computer-implemented method of claim 11, comprising identifying a non-technical loss associated with the MD based on verification of the reported value indicating the reported value exceeds the measured value.

20. The computer-implemented method of claim 11, comprising assigning the class to the MD based on one or more properties of a client account that corresponds to the MD.

21. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, the computer-program product including instructions operable to cause a processor to perform operations comprising:

identify direct measurement device data comprising a measured value generated by a measurement device (MD), the measured value received as part of a first data stream comprising a series of measured values over an interval in a time domain, the first data stream communicated via an event stream processing (ESP) interface of the measurement device;

identify indirect MD data comprising a reported value associated with generation by the MD, the reported value and a unique identifier of the MD received as part of a second data stream comprising a series of reported values over the interval in the time domain and the unique identifier of the MD, the series of reported values associated with generation by the MD in a staging area based on the unique identifier received in the second data stream, the staging area communicatively coupled with the MD via a neighborhood area network (NAN) and the staging area comprising a temporary storage to aggregate reported values for a group of MDs associated with a geographic area of the MD;

generate an expected value for the reported value with a first machine learning algorithm based on a class assigned to the MD and a portion of the series of reported values received in the second data stream prior to the reported value being received;

compare the reported value to the expected value and the reported value to the measured value to validate the reported value associated with generation by the MD in the staging area communicatively coupled to the MD via the NAN;

transform at least one of the direct MD data and the indirect MD data into the frequency domain to generate a spectral density for the MD;

compare the spectral density for the MD with a model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD, wherein the model spectral density is generated by a second machine learning algorithm with data from a set of MDs identified to characterize the class assigned to the MD;

identify a technical loss or a non-technical loss associated with the MD based on one or more of the validation of the reported value and the verification of the class associated with the MD; and

determine one or more corrective processes associated with the MD to perform based on whether technical loss or non-technical loss associated with the MD is identified.

22. The computer-program product of claim 21, including instructions operable to cause the processor to perform operations comprising decomposing the at least one of the direct MD data and the indirect MD data into a summation of periodic functions to transform the at least one of the direct MD data and the indirect MD data into the frequency domain to generate the spectral density for the MD.

23. The computer-program product of claim 22, including instructions operable to cause the processor to perform operations comprising:

correlate an equipment type with at least one of the periodic functions included in the summation; and

utilize the equipment type to generate the expected value for the reported value.

24. The computer-program product of claim 22, including instructions operable to cause the processor to perform operations to compare of the spectral density for the MD with the model spectral density that corresponds to the class assigned to the MD to verify the class assigned to the MD comprising:

perform clustering with correlation analysis or principal component analysis of the spectral density for the MD and a plurality of additional spectral densities for the set of MDs identified to characterize the class assigned to the MD; and

determine whether the MD has similar frequency content similar to one or more of the set of MDs identified to characterize the class assigned to the MD based on the clustering.

25. The computer-program product of claim 21, wherein the reported value is aggregated in a second staging area communicatively coupled to the staging area in the NAN by a regional area network (RAN) prior to receival as part of the second live steam, the second staging area comprising a temporary storage to aggregate reported values for a second group of MDs associated with a geographic region of the MD, the geographic region of the MD comprising the geographic area of the MD and a second geographic area of at least one other MD.

26. The computer-program product of claim 21, including instructions operable to cause the processor to perform operations comprising combine the first data stream and the second data stream into a data record stream associated with the MD, wherein the data record stream is provided to a database for storage, and wherein the first and second machine learning algorithms are trained with data stored in the database as part of a plurality of data record streams associated with a plurality of MDs.

27. The computer-program product of claim 26, wherein the data from the set of MDs identified to characterize the class assigned to the MD used by the second machine learning algorithm to generate the model spectral density was stored in the database as part of a respective data record streams corresponding to each of the set of MDs identified to characterize the class assigned to the MD.

28. The computer-program product of claim 26, wherein training at least one of the first and second machine learning algorithms utilizes one or more of support vector machines (SVMs), Fourier transforms, logistic regression, deep neural networks (DNNs), principal component analysis (PCA), clustering in the frequency domain.

29. The computer-program product of claim 21, including instructions operable to cause the processor to perform operations comprising identify a non-technical loss associated with the MD based on verification of the reported value indicating the reported value exceeds the measured value.

30. The computer-program product of claim 21, including instructions operable to cause the processor to perform operations comprising assign the class to the MD based on one or more properties of a client account that corresponds to the MD.