Market-adaptive detection of plug-in electric vehicle charging using whole house energy metering data
The invention provides improved methods for detecting the presence of a plug-in electric vehicle (PEV) at a location based on interval measurements from whole house electrical meters. This methodology is applicable to multiple detection algorithms and has been validated using actual PEV customer data.
1. A computer-implemented method for detecting plug-in electrical vehicle (PEV) charging at a whole house electrical meter location, comprising the steps of:
providing a plurality of whole house electrical meters, wherein each whole house electrical meter is located at a particular whole house electrical meter location, and provides input data regarding that particular whole house electrical meter location;
retrieving the input data from a tangible computer memory, said input data including (i) whole house meter interval data for a single whole house meter location to be tested and (ii) whole house meter interval data for a superset of whole house meter locations;
calculating time-referenced summary statistics comprising one or both of the means and standard deviations for the whole house meter interval data for the superset;
selecting a time period for testing a location for PEV charging;
generating a SINGLE HOME MATRIX which is a matrix of whole house meter interval data to be tested;
generating one or both of:
a MEAN MATRIX which is a matrix of corresponding interval means from the superset, and
a STANDARD DEVIATION MATRIX which is a matrix of corresponding interval standard deviations from the superset;
normalizing the SINGLE HOME MATRIX using one or both of the MEAN MATRIX and the STANDARD DEVIATION MATRIX to generate a NORMALIZED MATRIX;
applying a binary hypothesis test to the NORMALIZED MATRIX to obtain a determination of whether PEV charging is present at the single whole house meter location;
and
storing a record of the determination in tangible computer memory.
2. The computer-implemented method of claim 1 , wherein the superset of whole house meter locations is selected from the group consisting of the whole population of whole house meter locations.
3. The computer-implemented method of claim 1 , wherein the superset of whole house meter locations is a subgroup of the whole population of whole house meter locations.
4. The computer-implemented method of claim 3 , wherein the subgroup is a subgroup of the whole population of whole house meter locations based on one or more of geographical location, type of home and demographics.
5. The computer-implemented method of claim 1 , wherein the step of normalizing the SINGLE HOME MATRIX to generate a NORMALIZED matrix, comprises:
element-wise subtraction of the MEAN MATRIX from the SINGLE HOME MATRIX to obtain element-wise differences; and/or
element-wise division of the element-wise differences by the STANDARD DEVIATION MATRIX to obtain element-wise results that are assigned to the NORMALIZED MATRIX.
6. The computer-implemented method of claim 1 , wherein the interval data is hourly interval data.
7. The computer-implemented method of claim 1 , wherein the period selected to be tested is a 24-hour period.
8. The computer-implemented method of claim 1 , wherein the method is performed for each of a plurality of single whole house meter locations in the superset to be tested.
9. The computer-implemented method of claim 1 , wherein the detection algorithm is selected from the group consisting of the Likelihood Ratio Test and Support Vector Machine classification.
10. The computer-implemented method of claim 1 , further comprising the steps of:
for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location; and/or
performing a distribution infrastructure load assessment based at least in part on the detection of PEV charging at whole house meter locations according to the method.
11. The computer-implemented method of claim 10 , wherein the method comprises the step of:
for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location, and
the reduced billing rate is assigned only for a period in which PEV charging is detected according to the method.
12. A computer-implemented method for detecting plugin electrical vehicle (PEV) charging at a whole house electrical meter location using a Support Vector Machine (SVM) algorithm, comprising the steps of:
providing a plurality of whole house electrical meters, wherein each whole house electrical meter is located at a particular whole house electrical meter location, and provides input data regarding that particular whole house electrical meter location;
retrieving the input data from a tangible computer memory, said input data including (i) whole house meter interval data for a single whole house meter location to be tested and (ii) whole house meter interval data for a superset of whole house meter locations;
calculating time-referenced summary statistics comprising the means for the whole house meter interval data for the superset;
selecting a time period for testing a location for PEV charging; generating a SINGLE HOME MATRIX which is a matrix of whole house meter interval data to be tested (x);
generating a TEMPORAL SEGMENTATION MATRIX which is a matrix of corresponding interval means from the superset (n);
selecting a NOMINAL DAY MODEL MATRIX of whole house meter interval data from a set of parameterized matrices with no PEV charging based on what matrix has the smallest norm (i.e. Euclidean/L2 norm) of difference with respect to the TEMPORAL SEGMENTATION MATRIX (so NOMINAL DAY MODEL MATRIX*
H=TEMPORAL SEGMENTATION MATRIX, but for an infinite set of matrices, NOMINAL DAY MODEL MATRIX=\i=TEMPORAL SEGMENTATION MATRIX);
generating a PEV DAY MODEL MATRIX of whole house meter interval data that is the element-wise sum of the NOMINAL DAY MODEL MATRIX plus a predetermined sequence of PEV charging load that is typical for the time period selected (n+Xi); calculating the norm for the difference of SINGLE HOME MATRIX minus NOMINAL DAY MODELMATRIX (i.e. Euclidian/L2 norm) and then taking the reciprocal, assigned to NULL (=1/∥x−\i∥);
calculating the norm for the difference of SINGLE HOME MATRIX minus PEV DAY MODELMATRIX (i.e. Euclidian/L2 norm) and then taking the reciprocal, assigned to ALTERNATE (=1/1|x−(n+Xi)∥=1/∥x−n−Xi∥); dividing ALTERNATE by NULL to obtain a RATIO value; making the determination that if the RATIO value is greater than a preselected calibration parameter, then a PEV is detected, else no PEV is detected for the selected single whole house meter location during the time period being tested; and storing a record of the determination in tangible computer memory.
13. The computer-implemented method of claim 12 , wherein the superset of whole house meter locations is selected from the group consisting of the whole population of whole house meter locations.
14. The computer-implemented method of claim 12 , wherein the superset of whole house meter locations is a subgroup of the whole population of whole house meter locations.
15. The computer-implemented method of claim 14 , wherein the subgroup is a subgroup of the whole population of whole house meter locations based on one or more of geographical location, type of home and demographics.
16. The computer-implemented method of claim 12 , wherein the interval data is hourly interval data.
17. The computer-implemented method of claim 12 , wherein the period selected to be tested is a 24-hour period.
18. The computer-implemented method of claim 12 , wherein the method is performed for each of a plurality of single whole house meter locations in the superset to be tested.
19. The computer-implemented method of claim 12 , further comprising the step of:
for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location; and/or
performing a distribution infrastructure load assessment based at least in part on the detection of PEV charging at whole house meter locations according to the method.
20. The computer-implemented method of claim 19 , wherein the method comprises the step of: for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location, and
the reduced billing rate is assigned only for a period in which PEV charging is detected according to the method.
21. A computer-implemented method for detecting plugin electrical vehicle (PEV) charging at a whole house electrical meter location using a Likelihood Ration Test (LRT) algorithm, comprising the steps of:
providing a plurality of whole house electrical meters, wherein each whole house electrical meter is located at a particular whole house electrical meter location, and provides input data regarding that particular whole house electrical meter location;
retrieving the input data from a tangible computer memory, said input data including (i) whole house meter interval data for a single whole house meter location to be tested and (ii) whole house meter interval data for a superset of whole house meter locations;
calculating time-referenced summary statistics comprising the means and standard deviations for the whole house meter interval data for the superset; selecting a time period for testing a location for PEV charging; generating a SINGLE HOME MATRIX which is a matrix of whole house meter interval data to be tested (x);
generating one or both of:
a MEAN MATRIX which is a matrix of corresponding interval means from the superset (\i), and
a STANDARD DEVIATION MATRIX which is a matrix of corresponding interval standard deviations from the superset (a);
generating a TEMPORAL SEGMENTATION MATRIX which is a concatenated matrix of MEAN MATRIX and STANDARD DEVIATION MATRIX [\i, a];
selecting a joint probability density/mass function from a set of parameterized functions for no PEV charging (null function) based on which function's corresponding matrix of concatenated means and standard deviations has the smallest norm (i.e. Euclidean/L2 norm) of difference with respect TEMPORAL SEGMENTATION MATRIX;
generating a joint probability density/mass function for a PEV charging (alternate function) by adding a mean and standard deviation adjustment to the null function which could be as simple as a constant mean offset by charge rate R or a time-weighted offset for mean and standard deviation;
calculating the alternate function value for the SINGLE HOME MATRIX and assign to ALTERNATE HYPOTHESIS;
dividing the ALTERNATE HYPOTHESIS by the NULL HYPOTHESIS to obtain a RATIO value; and
making the determination that if the RATIO value is greater than a preselected calibration parameter, then a PEV is detected, else no PEV is detected for the selected whole house meter location during the time period being tested; and
storing a record of the determination in tangible computer memory.
22. The computer-implemented method of claim 21 , wherein the alternate function is
Alternate Hypothesis: Θ 1,k ( R )=[μ k +R,σ k ,A k ].
23. The computer-implemented method of claim 21 , wherein the alternate function is
f
daily
(
x
❘
Θ
1
(
R
)
)
≈
∑
(
e
)
∑
(
s
)
f
daily
(
x
❘
Θ
1
(
R
)
,
s
,
e
)
f
s
,
R
(
s
)
f
e
,
R
(
e
)
.
24. The computer-implemented method of claim 21 , wherein the superset of whole house meter locations is selected from the group consisting of the whole population of whole house meter locations.
25. The computer-implemented method of claim 21 , wherein the superset of whole house meter locations is a subgroup of the whole population of whole house meter locations.
26. The computer-implemented method of claim 25 , wherein the subgroup is a subgroup of the whole population of whole house meter locations based on one or more of geographical location, type of home and demographics.
27. The computer-implemented method of claim 21 , wherein the interval data is hourly interval data.
28. The computer-implemented method of claim 21 , wherein the period selected to be tested is a 24-hour period.
29. The computer-implemented method of claim 21 , wherein the method is performed for each of a plurality of single whole house meter locations in the superset to be tested.
30. The computer-implemented method of claim 21 , further comprising the step of:
for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location; and/or
performing a distribution infrastructure load assessment based at least in part on the detection of PEV charging at whole house meter locations according to the method.
31. The computer-implemented method of claim 21 , wherein the method comprises the step of:
for a whole house meter location for which a determination is made that PEV charging is present, assigning a reduced billing rate for at least some electrical power consumption at that whole house meter location, and
the reduced billing rate is assigned only for a period in which PEV charging is detected according to the method.