Patent Grant
9389112
This claims the benefit of German Patent Application DE 10 2011 122 127.5, filed Dec. 22, 2011 and hereby incorporated by reference herein.
The present invention relates to a method and a device for detecting the tank filling level in a motor vehicle.
Such approaches work by detecting an analog signal of a tank filling level sensor. The signal is sharply filtered through algorithms implemented in a control unit in order to reduce the sloshing in the signal caused by driving. Since the characteristic constant of the low-pass filter must be large, the low-pass filter becomes slower. The filtered signal will not indicate the prevailing filling level but instead will indicate a previous filling level. This is noticeable in particular when the liquid is being pumped out of the tank, as in the case of a fuel tank or an ADBLUE (a trademark as described below, also known as a diesel exhaust fluid or AUS 32) tank.
In addition, there are complex methods which attempt to increase the reliability of the low-pass filter result by taking into account the vehicle state (acceleration, braking, maintaining speed, stopping, etc.). Other methods use a second low-pass filter, which is faster than the first. This second low-pass filter should supply more up-to-date information about the filling level but its area of use is limited to vehicle states in which the sloshing turns out to be lower.
Other methods subtract the accumulated spent setpoint quantity of fuel in the case of a vehicle tank or the accumulated metered setpoint ADBLUE quantity in the case of an ADBLUE tank from an initial filling level during the sloshing phase to ascertain the filling level. The initial filling level should be measured by the filling level sensor in a “resting” phase. Each time the vehicle is in a “resting” phase, the initial filling level is measured again and the accumulated setpoint quantity is set to zero. This method relates to the desired quantity or the quantity to be injected or the quantity to be metered since the actual quantity injected is unknown due to tolerances, drift, aging, etc. This results in a deviation in the actual tank filling level, which increases over time and may be reset in the next “resting” phase.
Resetting has as a reference the tank filling level sensor value, which may be a possible source of error with certain tank filling level sensors, e.g., with so-called switching stage sensors. This method would be at fault if, for example, the tank filling level measured by the filling level sensor corresponds to a certain switching stage but the actual filling level is closer to the next stage.
The known methods have the disadvantage that additional characteristic variables and safety measures are required, resulting in substantially greater effort in calibration and in tests.
It is an object of the present invention to provide a device and a method for detecting the tank filling level in a motor vehicle, which avoid the disadvantages mentioned above while being simpler, more accurate and more reliable.
The method described here and the corresponding device ascertain the tank filling level continuously (independently of the type of tank filling level sensor, the switching stage, ultrasound, etc.) as a function of the mathematical relationship between the tank volume and the tank filling level, from the setpoint quantity injected or metered and pumped out of the tank, from the stochastic properties of sloshing and from the stochastic scattering of the fuel injector or the metering module.
With this information, the tank filling level system may be modeled mathematically with the aid of a state space system, and the tank volume value as well as the tank filling level may be ascertained continuously and the tank filling level sensor value may be filtered by using a Kalman filter.
According to the present invention, it is advantageous here that a more detailed method of ascertaining the tank filling level and the tank volume is possible.
A few advantages are described in keywords as follows:
The present invention will now be presented on the basis of an exemplary embodiment with reference to the drawings, in which:
The use of a Kalman filter to ascertain the tank filling level value is illustrated below for use in a system such as that shown in
The Kalman filter algorithm relates to the estimation of a time-discrete process, which may be represented by the following differential equation:
Xk=Fk-1Xk-1+Bk-1uk-1+wk-1 (1)
Zk=HkXk+vk (2)
Xk: the system state variable to be estimated.
Fk-1: describes the transitions between states Xk-1 and Xk which follow one another chronologically.
Bk-1: describes the relationship between Xk and uk-1.
uk-1: is the active disturbance.
wk-1: is the noise term and represents the random components that cannot be detected. The noise wk-1 not correlated over time has a mean value of 0 and a covariance Q.
Zk: is the measured variable.
Hk: describes the relationship between measured variable Zk and state variable Xk.
vk: represents the measurement noise; “vk” has a mean value of 0 and a covariance R.
The Kalman filter algorithm estimates state variable Xk on the basis of a recursive method using a predictive component:
{circumflex over (x)}k|k-1=Fk-1{circumflex over (x)}k-1+Bk-1uk-1 (3)
{circumflex over (P)}k|k-1=Fk-1{circumflex over (P)}k-1Fk-1T+Qk-1 (4)
and a corrective component:
{circumflex over (x)}k={circumflex over (x)}k|k-1+{circumflex over (K)}k{tilde over (y)}k (5)
{circumflex over (P)}k={circumflex over (P)}k|k-1−{circumflex over (K)}kSk{circumflex over (K)}kT (6)
{tilde over (y)}k=zk−Hk{circumflex over (x)}k|k-1 (7)
Sk=Hk{circumflex over (P)}k|k-1HkT+Rk (8)
{circumflex over (K)}k={circumflex over (P)}k|k-1HkTSk−1 (9)
The present invention relates to the application of the Kalman filter for estimating a tank filling level. First, the different state variables must be identified.
In an ideal tank having an ideal filling level sensor, i.e., tolerances are equal to 0, and there is no sloshing, the filling level may be described mathematically:
L=h*Vol (10)
L: measured level
Vol: volume
h: is the mathematical relationship between level and volume; “h” is defined by the geometry of the tank, e.g., for a cube-shaped container, h=1/A: A=area perpendicular to the sensor.
Another mathematical equation which holds for this system is:
Vol=Vol0−d*m (11)
where
Vol: volume
Vol0: volume of the tank at time=0
d: liquid density
m: mass to be injected
In a nonideal situation—the vehicle is moving on an undulating road, for example—the tolerances of the filling level sensor and the injection module as well as sloshing must be taken into account.
Sloshing and the filling level sensor tolerance are added in equation (10):
L=h*Vol+s (12)
where “s” denotes the random and not detectable components of filling level sensor tolerance and sloshing; “s” is a stochastic variable having a mean value of 0 and a known covariance Q.
The tolerance of the injection module (w) must be added in (11).
Vol=Vol0−d*m+w (13)
where “w” is a stochastic variable having mean value of 0 and a known covariance R.
Equations (12) and (13) may be rewritten as (1) and (2):
(13)→ (1)
Volk=(1)Volk-1+(−d)*mk-1+wk-1 (14)
(12)→ (2)
Lk=hVolk+sk (15)
for one computation cycle k
Xk=Volk
Fk-1=1
Bk-1=d
uk-1=mk-1
wk-1=w
Zk=Lk
Hk=h
vk-1=s
The prevailing volume of a container may be ascertained with the aid of a Kalman filter using these identified elements of the state space despite the interferences such as component tolerances and sloshing for each computation cycle k.
The diagram in
ADBLUE (ISO 22241/DIN 70070/AUS32) is the brand name of a clear synthetic 32.5% solution of high purity urea in demineralized water, which is used for after treatment of exhaust gas in an SCR catalytic converter. By selective catalytic reduction (SCR), the emission of nitrogen oxides (NOx) is reduced by about 90% (in steady-state operation). The German Association of the Automobile Industry (VDA) holds the trademark rights for ADBLUE. Filling level sensor (2) detects the filling level via a floating gauge (3). Sloshing (4) caused by the movement of the vehicle interferes with correct detection of the filling level. The filling level sensor value is transmitted to control unit (8) over electrical line (9). The control unit controls pump (5) via electrical line (10) and controls metering module (6) via electrical line (11). Pump (5) pumps ADBLUE through tubes (12) and (13) in the direction of metering module (6), and metering module (6) injects ADBLUE into the SCR catalytic converter (7).
The control unit calculates the quantity of ADBLUE to be metered as a function of external influencing factors. The control unit must monitor the tank filling level and issue warnings, if necessary, and in the extreme case must reduce engine power, among other possible responses.
The Kalman filter is illustrated in
for each computation cycle k:
Lk is the noise level measured by filling level sensor (2).
mk-1 is the setpoint quantity to be injected by metering module (6).
Volk is the volume ascertained and filtered by the Kalman filter.
The graph in
The blue line, resembling an envelope curve, shows the volume detected by filling level sensor (2).
The yellow line (bottom) indicates the volume ascertained with the aid of formula (11).
The red line (top) indicates the volume ascertained with the aid of the Kalman filter according to the present invention.
The present invention is suitable for ascertaining the filling level of systems, which include a tank, a pump and a fuel injector. The present invention is suitable in particular for systems in moving vehicles.
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2011 122 127 | Dec 2011 | DE | national |
| Number | Name | Date | Kind |
|---|---|---|---|
| 6157894 | Hess | Dec 2000 | A |
| 6502042 | Eid | Dec 2002 | B1 |
| 7603242 | Tichborne et al. | Oct 2009 | B2 |
| 20040073386 | Benedetti | Apr 2004 | A1 |
| 20040079150 | Breed | Apr 2004 | A1 |
| 20050072226 | Pappas | Apr 2005 | A1 |
| 20060025897 | Shostak | Feb 2006 | A1 |
| 20080236275 | Breed | Oct 2008 | A1 |
| 20130111987 | Ciani | May 2013 | A1 |
| Number | Date | Country |
|---|---|---|
| 19929295 | Dec 2000 | DE |
| Number | Date | Country | |
|---|---|---|---|
| 20130160541 A1 | Jun 2013 | US |
