Method and apparatus for starter motor diagnosis and prognosis using parameter estimation algorithm
A system and method for determining the state of health of a starter motor to notify a vehicle driver of a potential starter motor failure before the failure actually occurs. The starter motor includes an armature and motor brushes each providing a resistance, and an armature coil providing an armature inductance. Further, the starter motor has a back EMF because of the starter motor being coupled to a flywheel and the vehicle engine. The system and method monitor the combined resistance of the armature and the motor brushes, the inductance of the armature and a back EMF constant of the motor, and provide a signal indicating a potential starter motor failure if any of these three values significantly deviates from nominal values. In one embodiment, the analysis of the motor resistance, armature inductance and back EMF constant is provided by a regression model to determine estimated motor parameters.
1. A method for determining the state of health of a starter motor in a vehicle, said method comprising:
using a current sensor to determine a starter motor current of the starter motor;
using a voltage sensor to determine a starter motor voltage of the starter motor;
providing an engine speed of an engine of the vehicle;
identifying starter motor values that will be used to determine the state of health of the starter motor, wherein the starter motor values are a starter motor resistance that is the combination of a starter motor armature resistance and a starter motor brush resistance, a starter motor armature inductance and a back EMF motor constant;
identifying a time model of the starter motor using the starter motor voltage, the starter motor current, the engine speed and the starter motor values;
defining model parameters from the time model;
providing a regression model based on the time model;
using a controller to determine motor parameters based on the regression model;
using the controller to determine an error using the motor parameters; and
using the controller to compare the error to an error threshold to determine the state of health of the starter motor.
2. The method according to claim 1 wherein identifying a time model of the starter motor includes using the equation:
L
m
ⅆ
I
a
ⅆ
t
=
-
R
m
I
a
+
V
m
-
K
m
T
S
ω
E
where L m is the armature inductance, I a is the starter motor current, R m is the starter motor resistance, V m is the starter motor voltage, K m is the back-EMF motor constant, T s is a gear ratio between a starter motor shaft and an engine shaft and ω E is the engine speed.
3. The method according to claim 2 wherein defining model parameters from the time model includes defining model parameters p 1 , p 2 and p 3 as:
p
1
=
(
1
-
Δ
tR
m
L
m
)
p
2
=
Δ
t
L
m
p
3
=
Δ
t
L
m
K
m
T
S
where Δt is the sampling time.
4. The method according to claim 1 wherein providing a regression model includes using a recursive least squares algorithm with exponential forgetting.
5. The method according to claim 4 wherein the regression model uses the equations:
y ( t )=φ T ( k )θ
where y(k)=I a (k), φ T (k)={I a (k−1)V m (k−1),−ω E (k−1)}, y and φ are regression model values, k is a sample, I a is the starter motor current, V m is the starter motor voltage, ω E is the engine speed and θ represents model parameters p 1 , p 2 and p 3 .
6. The method according to claim 1 wherein providing a regression model includes using a batch least squares algorithm.
7. The method according to claim 6 wherein the regression model uses the equations:
Y
=
Φθ
Y
=
{
I
a
(
t
2
)
I
a
(
t
3
)
⋮
I
a
(
t
n
)
}
Φ
[
I
a
(
t
1
)
V
m
(
t
1
)
-
ω
E
(
t
1
)
I
a
(
t
2
)
V
m
(
t
2
)
-
ω
E
(
t
2
)
⋮
⋮
⋮
I
a
(
t
n
-
1
)
V
m
(
t
n
-
1
)
-
ω
E
(
t
n
-
1
)
]
θ
^
=
[
Φ
T
Φ
]
-
1
Φ
T
Y
where Y and Φ are regression model values, I a is the started motor current, V m is the starter motor voltage, ω E is the engine speed and θ is the model parameters.
8. The method according to claim 1 wherein the motor parameters R m , K m and L m are defined by the equations:
R
m
=
p
1
-
1
p
2
K
m
=
p
3
p
2
L
m
=
p
2
Δ
t
where p 1 , p 2 and p 3 are the model parameters and Δt is the sampling time.
9. The method according to claim 1 wherein the motor parameters are stored as a temperature in a look-up table.
10. The method according to claim 9 wherein determining an error includes determining an error using the equation:
e
=
max
{
w
R
R
^
m
-
K
m
O
(
T
)
R
m
O
(
T
)
,
w
K
K
⋒
m
-
K
m
O
(
T
)
K
⋒
m
,
w
L
L
^
m
-
L
m
O
(
T
)
L
^
m
}
where R m is a resistance motor parameter, K m is a back EMF motor constant parameter, L m is an inductance motor parameter, w R is a rating factor for the resistance motor parameter R m , W K is a rating factor for the back EMF constant motor parameter K m , w L , is a rating factor for the inductance motor parameter L m , K m o is the back EMF motor constant parameter based on temperature, R m o is the resistance motor parameter based on temperature and L m o is the inductance motor parameter based on temperature.
11. A method for determining the state of health of a starter motor in a vehicle, said method comprising:
using a current sensor to determine a starter motor current of the starter motor;
using a voltage sensor to determine a starter motor voltage of the starter motor;
providing an engine speed of an engine of the vehicle;
identifying a starter motor resistance that is a combination of a starter motor armature resistance and a starter motor brush resistance;
identifying a starter motor armature inductance;
identifying a back EMF motor constant;
identifying a time model of the starter motor using the starter motor voltage, the starter motor current, the engine speed, the starter motor resistance, the starter motor armature inductance and the back EMF motor constant;
defining model parameters from the time model;
providing a regression model based on the time model;
using a controller to determine motor parameters R m , K m and L m as
R
m
=
p
1
-
1
p
2
,
K
m
=
p
3
p
2
and
L
m
=
p
2
Δ
t
based on the regression model, where p 1 , p 2 and p 3 are the model parameters, R m is a resistance motor parameter, K m is a back EMF constant motor parameter and L m is an inductance motor parameter;
using the controller to determine an error using the motor parameters; and
using the controller to compare the error to an error threshold to determine the state of the health of the starter motor.
12. The method according to claim 11 wherein identifying a time model of the starter motor includes using the equation:
L
m
ⅆ
I
a
ⅆ
t
=
-
R
m
I
a
+
V
m
-
K
m
T
S
ω
E
where I a is the starter motor current, V m is the starter motor voltage, T s is a gear ratio between a starter motor shaft and an engine shaft and ω E is the engine speed.
13. The method according to claim 11 wherein providing a regression model includes using a recursive least squares algorithm with exponential forgetting.
14. The method according to claim 11 wherein providing a regression model includes using a batch least squares algorithm.
15. A method for determining the state of health of a starter motor in a vehicle, said method comprising:
using sensors to determine a starter motor voltage and a starter motor current of the starter motor;
identifying a time model of the starter motor using a the starter motor voltage, a the starter motor current, an engine speed, a starter motor resistance, a starter motor armature inductance and a back EMF motor constant;
defining model parameters from the time model; and
using the model parameters in a controller to determine an error that is used to determine the state of health of the starter motor.
16. The method according to claim 15 further comprising providing a regression model based on the time model, said regression model being used to convert the model parameters to the error.
17. The method according to claim 16 wherein providing a regression model includes using a recursive least squares algorithm with exponential forgetting.
18. The method according to claim 16 wherein providing a regression model includes using a batch least squares algorithm.
19. The method according to claim 15 wherein identifying a time model of the starter motor includes using the equation:
L
m
ⅆ
I
a
ⅆ
t
=
-
R
m
I
a
+
V
m
-
K
m
T
S
ω
E
where L m is the armature inductance, I a is the starter motor current, R m is the starter motor resistance, V m is the starter motor voltage, K m is the back-EMF motor constant, T s is a gear ratio between a starter motor shaft and an engine shaft and ω E is the engine speed.
20. A method for determining the state of health of a starter motor in a vehicle, said method comprising:
using a current sensor to determine a starter motor current of the starter motor;
using a voltage sensor to determine a starter motor voltage of the starter motor;
providing an engine speed of an engine of the vehicle;
identifying starter motor values that will be used to determine the state of health of the starter motor;
identifying a time model of the starter motor using the starter motor voltage, the starter motor current, the engine speed and the starter motor values;
defining model parameters from the time model;
providing a regression model based on the time model;
using a controller to determine motor parameters based on the regression model, wherein the motor parameters are stored as a temperature in a look-up table;
using the controller to determine an error using the motor parameters; and
using the controller to compare the error to an error threshold to determine the state of health of the starter motor.