Battery diagnosis device, battery pack, battery system, and battery diagnosis method
A battery diagnosis device according includes a memory configured to store an observation matrix including a plurality of observation voltage vectors indicating a time-series of cell voltage of each of battery cells, and a controller configured to determine a plurality of principal component vectors, a plurality of singular values and a plurality of coefficient vectors from the observation matrix. Each coefficient vector includes a plurality of coefficients corresponding to the plurality of observation voltage vectors in a one-to-one relationship. The controller, for each coefficient vector, determines an invalid coefficient among the plurality of coefficients by comparing the plurality of coefficients included in the corresponding coefficient vector, and detects abnormality of the battery cell corresponding to the invalid coefficient based on the principal component vector corresponding to the corresponding coefficient vector, the singular value corresponding to the corresponding coefficient vector and the invalid coefficient.
1 . A battery diagnosis device, comprising:
a memory configured to store an observation matrix including a plurality of observation voltage vectors indicating a time-series of a cell voltage of each of a plurality of battery cells; and
a controller configured to:
determine a plurality of principal component vectors, a plurality of singular values and a plurality of coefficient vectors from the observation matrix, wherein each coefficient vector includes a plurality of coefficients corresponding to the plurality of observation voltage vectors in a one-to-one relationship, and
for each coefficient vector:
determine an invalid coefficient among the plurality of coefficients by comparing the plurality of coefficients included in the corresponding coefficient vector, and
detect abnormality of a battery cell corresponding to the invalid coefficient among the plurality of battery cells by following a diagnosis procedure in sequence of:
determining a principal component vector corresponding to the corresponding coefficient vector among the plurality of principal component vectors,
determining a singular value corresponding to the corresponding coefficient vector among the plurality of singular values, and
determining the invalid coefficient; and
detect the battery cell corresponding to the invalid coefficient among the plurality of battery cells as faulty when a voltage characteristic value of the partial voltage vector is larger than a second reference value,
wherein the controller is further configured to activate a predetermined protection operation when the battery cell is detected as faulty, and
wherein the predetermined protection operation includes turning off a switch.
2 . The battery diagnosis device according to claim 1 , wherein the controller is further configured to determine a first sub-matrix, a second sub-matrix and a third sub-matrix by applying a matrix decomposition algorithm to the observation matrix,
the first sub-matrix includes the plurality of principal component vectors as column vectors,
the second sub-matrix includes the plurality of singular values as elements of a principal diagonal, and
the third sub-matrix includes the plurality of coefficient vectors as row vectors.
3 . The battery diagnosis device according to claim 1 , wherein the controller is further configured to determine, as the invalid coefficient, a coefficient of which an absolute value of a difference between the coefficient and an average of the plurality of coefficients among the plurality of coefficients is larger than a first reference value.
4 . The battery diagnosis device according to claim 3 , wherein the controller is further configured to determine the first reference value by multiplying a standard deviation of the plurality of coefficients by a first scaling factor.
5 . The battery diagnosis device according to claim 1 , wherein the controller is further configured to, for each coefficient vector:
extract a partial voltage vector of an observation voltage vector corresponding to the invalid coefficient among the plurality of observation voltage vectors by multiplying the principal component vector corresponding to the corresponding coefficient vector among the plurality of principal component vectors, the singular value corresponding to the corresponding coefficient vector among the plurality of singular values and the invalid coefficient.
6 . The battery diagnosis device according to claim 5 , wherein the controller is further configured to determine the voltage characteristic value to be equal to a difference between a maximum partial voltage and a minimum partial voltage among a plurality of partial voltages included in the partial voltage vector.
7 . The battery diagnosis device according to claim 5 , wherein the controller is further configured to determine the second reference value to be equal to a value obtained by multiplying a voltage resolution of a voltage measurement circuit by a second scaling factor.
8 . The battery diagnosis device according to claim 1 , wherein the controller is further configured to output a fault message when a ratio of a maximum singular value to a minimum singular value among the plurality of singular values is less than a preset value.
9 . A battery pack comprising the battery diagnosis device according to claim 1 .
10 . A battery system comprising the battery pack according to claim 9 .
11 . The battery diagnosis device according to claim 1 , wherein the abnormality of the battery cell is detected based on only the cell voltage from among a current, a temperature and the cell voltage of the battery cell.
12 . A battery diagnosis method, comprising:
determining a plurality of principal component vectors, a plurality of singular values and a plurality of coefficient vectors from an observation matrix including a plurality of observation voltage vectors indicating a time series of a cell voltage of each of a plurality of battery cells,
wherein each coefficient vector includes a plurality of coefficients corresponding to the plurality of observation voltage vectors in a one-to-one relationship;
for each coefficient vector:
determining an invalid coefficient among the plurality of coefficients by comparing the plurality of coefficients included in the corresponding coefficient vector;
detecting abnormality of a battery cell corresponding to the invalid coefficient among the plurality of battery cells by following a diagnosis procedure in sequence of:
determining a principal component vector corresponding to the corresponding coefficient vector among the plurality of principal component vectors,
determining a singular value corresponding to the corresponding coefficient vector among the plurality of singular values, and
determining the invalid coefficient;
detecting the battery cell corresponding to the invalid coefficient among the plurality of battery cells as faulty when a voltage characteristic value of the partial voltage vector is larger than a second reference value; and
activating a predetermined protection operation when the battery cell is detected as faulty,
wherein the predetermined protection operation includes turning off a switch.
13 . The battery diagnosis method according to claim 12 , wherein the determining the invalid coefficient among the plurality of coefficients comprises determining, as the invalid coefficient, a coefficient of which an absolute value of a difference between the coefficient and an average of the plurality of coefficients among the plurality of coefficients is larger than a first reference value.
14 . The battery diagnosis method according to claim 12 , wherein the detecting the abnormality of the battery cell corresponding to the invalid coefficient among the plurality of battery cells comprises:
extracting a partial voltage vector of an observation voltage vector corresponding to the invalid coefficient among the plurality of observation voltage vectors by multiplying the principal component vector corresponding to the corresponding coefficient vector among the plurality of principal component vectors, the singular value corresponding to the corresponding coefficient vector among the plurality of singular values and the invalid coefficient.
15 . The battery diagnosis device according to claim 1 , wherein the controller detects the abnormality of the battery cell based on only the cell voltage of the battery cell among the plurality of battery cells of the observation matrix.
16 . The battery diagnosis device according to claim 1 , wherein the time-series of the cell voltage is obtained over a specified period of a time window that moves by a preset or adjustable time interval.
17 . The battery diagnosis device according to claim 1 , wherein the predetermined protection operation further includes outputting a diagnosis message.
18 . The battery diagnosis method according to claim 12 , wherein the detecting the abnormality of the battery cell is based on only the cell voltage of the battery cell among the plurality of battery cells of the observation matrix.
19 . The battery diagnosis method according to claim 12 , wherein a time-series of the cell voltage is obtained over a specified period of a time window that moves by a preset or adjustable time interval.
20 . The battery diagnosis method according to claim 12 , wherein the abnormality of the battery cell is detected based on only the cell voltage from among a current, a temperature and the cell voltage of the battery cell.