System and method for determining vehicle component conditions via performance correlation
A method for determining vehicle component conditions via performance correlation is provided. A list of doors for maintenance on a transport vehicle is maintained. Measurements for one of the doors based on an inspection of that door are maintained. A determination is made as to whether maintenance is required for the door based on the measurements and a maintenance status is assigned to the door. The door measurements are compared to measurements for other doors of the transportation vehicle. Those other doors with measurements similar to the door are identified and the maintenance status of the door is assigned to the other doors identified.
1. A method for determining vehicle component conditions via performance correlation, comprising:
maintaining a list of doors on a transport vehicle;
receiving measurements for one of the doors based on an inspection of that door;
determining whether maintenance is required for the door based on the measurements and assigning a maintenance status to the door comprising one of maintenance required and no maintenance required;
identifying a portion of other doors of the transport vehicle with measurements similar to the door, comprising:
inputting the measurements of the door and each of the other doors into a classifier that is trained to determine similarity and dissimilarity of doors; and
determining one of a similarity and dissimilarity of each of the other doors to the door; and
identifying those doors that require maintenance without inspection of all the doors on the list, comprising:
applying the maintenance status assigned to the door to one or more of the other doors when the door and the one or more other doors are similar; and
removing the doors from the list with the maintenance status of no maintenance required.
2. A method according to claim 1 , further comprising:
when the maintenance status identifies that maintenance is required, maintaining the doors on the list.
3. A method according to claim 1 , wherein the measurements comprise motor current measurements from which features associated with a fault are determined comprising one or more of a mean value of the motor current during a state of the door, a highest value of the motor current in that state, a lowest value of the motor current for that state, an amount of variation of the motor current from the mean value, a time span for the state, a mean of the motor current during opening of the door, a mean of the motor current during closing of the door, an average outside temperature, and an average temperature of the motor.
4. A method according to claim 1 , further comprising:
identifying those other doors with measurements similar to the door, comprising generating vectors for the door f s i and each other door f t i ; and
comparing the vectors for the door with the vectors of each of the other doors.
5. A method according to claim 4 , further comprising:
determining the vectors for each door by selecting the features for an open and close cycle for each hour and each day over a predetermined time.
6. A method according to claim 4 , further comprising:
performing the comparison between the vectors of the door f s i and the vectors of each of the other doors f t i by assigning a correlation value according to the following equation:
Corrst
(
hour
,
month
,
year
)
=
∑
i
=
1
n
(
f
s
i
-
f
_
s
i
)
(
f
t
i
-
f
_
t
i
)
∑
i
=
1
n
(
f
s
i
-
f
_
s
i
)
2
∑
i
=
1
n
(
f
t
i
-
f
_
t
i
)
2
where f s i and f t i are the means of the vectors for the door f s i and the other door f t i over a time period, respectively.
7. A method according to claim 6 , further comprising:
applying a threshold to the correlation values of the doors; and
designating those doors with correlation values that satisfy the threshold as the other doors with measurements similar to the door.
8. A method according to claim 7 , further comprising:
determining the threshold based on past historical measurements of the doors.
9. A method according to claim 6 , further comprising:
displaying the correlation values of the other doors via a graph; and
determining that the other doors are similar based on the displayed correlation values.