DE10162599B4 - Method and device for monitoring speed sensors for missing teeth - Google Patents

Abstract

Method for monitoring at least one speed sensor, which emits an output signal (S1), which consists of a sequence of pulses, whereby - at predeterminable times, a first variable (NF / Δt1) assigned to the speed sensor is determined, which with the number of signals in one pulses (NF) occurring at the respective time correlated time interval (Δt1) is correlated, the first variable (NF / Δt1) being compared with a calculated lower limit value (UG), with at least the first variable (NF / Δt1), which was determined at a previous point of time, and - to monitor the speed sensor, depending on the result of this comparison, a suspected error signal is generated, characterized in that a first suspected error signal is generated as soon as the value falls below the lower limit in at least one comparison Limit (UG) by the first variable assigned to the speed sensor (NF / Δt1) is set, after issuing the first suspected error signal for those speed sensors for which a first suspected error signal has been generated, it is determined whether the drop below the respective lower limit value is repeated at a later time, the speed sensor consisting of a pulse wheel (11, 12) and a transducer (10), - and the later point in time is characterized in that the impulse wheel has made one full revolution and - that the full revolution of the impulse wheel is determined by the fact that between the detection of a first drop below the lower limit value (UG) and the detection of a second drop below the lower limit value (UG) determines a second variable correlated with the number of pulses that occur, which must lie in a predeterminable interval for the detection of a full revolution, the width of the predeterminable interval being dependent on the current vehicle reference speed (vref ) depends is ngig.

Description

State of the art

The invention is based on a method for fault detection in speed sensors.

In the DE 199 61 504 A1 proposes a method for detecting speed errors. In this case, pulses and pulse pauses are generated by poles which generate speed signals and which are arranged on the circumference of a pole wheel. In order to determine a change in speed, pulse pauses are compared in a first step for equality within a first permissible tolerance range and in a second step for inequality in terms of their duration during a wheel revolution.

In the DE 44 44 408 A1 a method for detecting a malfunction of a speed sensor is proposed. This can be, for example, a wheel speed sensor in an ABS system, in which wheel speeds are determined on the basis of the signals from the speed sensor. Filtered wheel speeds and filtered wheel speed changes are determined from these.

By comparing filtered and unfiltered quantities as well as plausibility tests, sensor errors, in particular a failure of the sensor signal, are recognized.

From the DE 691 11 300 T2 a device and a method for recognizing errors in the output signals of a speed sensor are known, in particular of “drop-out” errors in speed sensors of the type with a rotatable toothed rotor that runs past a rotationally fixed sensor or stator element,

The DE 196 02 359 A1 discloses a method and a circuit arrangement for monitoring a speed sensor, which consists of a measuring sensor rotating in the form of a toothed disk or the like and a stationary measuring sensor. Short-term changes or dips in the sensor output signal, which are repeated after each full rotation of the wheel, are recorded and evaluated for fault detection.

The DE 197 49 791 A1 discloses an evaluation method for an output signal of a sensor device scanning a cyclical movement.

The features of the preambles of the independent claims go from DE DE 691 11 300 T2 out.

Advantages of the invention

The invention relates to a method and a device for monitoring speed sensors for missing teeth.

Many methods known from the prior art use time measurement when checking speed sensors. Characteristic times of the sequence of pulses emitted by a speed sensor are determined. This can be, for example, the time length of individual pulses or the time interval between two pulses. Such a method is, for example, from that mentioned at the beginning DE 199 61 504 A1 known.

In the method and the device according to the invention, complex measurements of different time intervals on given pulse sequences are eliminated.

• - zu vorgebbaren Zeitpunkten eine dem Drehzahlfühler zugeordnete erste Größe ermittelt wird, welche mit der Zahl der in einem mit dem jeweiligen Zeitpunkt korrelierten Zeitintervall auftretenden Impulse korreliert ist.

Erfindungsgemäß wird

• - die erste Große mit einem berechneten unteren Grenzwert verglichen, wobei in die Berechnung des unteren Grenzwerts wenigstens die erste Große, die zu einem vorhergehenden Zeitpunkt ermittelt worden ist, eingeht und
• - zur Überwachung des Drehzahlfuhlers abhangig vom Ergebnis dieses Vergleichs ein Fehlerverdachtssignal erzeugt.

This is a method for monitoring at least one speed sensor, which emits an output signal that consists of a sequence of pulses, wherein

• - At predeterminable times, a first variable assigned to the speed sensor is determined, which is correlated with the number of pulses occurring in a time interval correlated with the respective time.

According to the invention

• comparing the first variable with a calculated lower limit value, the calculation of the lower limit value including at least the first variable that was determined at a previous point in time and
• - To monitor the speed sensor, depending on the result of this comparison, a suspected error signal is generated.

It is furthermore advantageous that a first suspected error signal is generated as soon as at least one comparison determines that a lower limit is undershot by the first variable assigned to the speed sensor.

Another advantage is that after the first suspected error signal has been issued, the monitoring is concentrated on those speed sensors whose assigned first values indicate that the respective lower limit value has been undershot. This concentration on one wheel is not absolutely necessary, but it leads to resource savings (saving of RAM and / or ROM memory space). Of course, the further simultaneous monitoring of all four wheels would be possible even after a first suspected error signal has been issued.

Advantageously, the method and the device are characterized in that after generation of the first suspected error signal for those speed sensors for which a first suspected error signal has been generated, it is determined whether the drop below the respective lower limit value is repeated at a later point in time.

• - dass der Drehzahlfuhler aus einem Impulsrad und einem Aufnehmer besteht,
• - dass der spatere Zeitpunkt dadurch gekennzeichnet ist, dass das Impulsrad eine volle Umdrehung gemacht hat und
• - dass die volle Umdrehung des Impulsrades dadurch festgestellt wird, dass zwischen der Detektion eines erstmaligen Unterschreitens des unteren Grenzwertes und der Detektion eines zweitmaligen Unterschreitens des unteren Grenzwertes eine mit der Zahl der auftretenden Impulse korrelierte zweite Große ermittelt wird, welche zur Detektion einer vollen Umdrehung in einem vorgebbaren Intervall liegen muss.

An advantageous embodiment of the invention is characterized in that

• - that the speed sensor consists of a pulse wheel and a sensor,
• - That the later point in time is characterized by the fact that the pulse wheel has made a full revolution and
• - That the full rotation of the pulse wheel is determined by the fact that between the detection of falling below the lower limit value for the first time and the detection of falling below the lower limit value a second time, a second variable correlated with the number of pulses occurring is determined, which is used to detect a full rotation in a predefinable interval.

A high level of robustness and precision of the method, as well as a high level of recognizability of the error, is advantageously achieved in that the width of the predeterminable interval depends on the current vehicle reference speed (the vehicle reference speed is the slow speed of the vehicle).

In the sense of a simple evaluation, it is advantageous if the second variable correlated with the number of pulses occurring is the number of pulse edges.

list of figures

• 1 zeigt den prinzipiellen Aufbau eines Drehzahlfuhlers und der zugehörigen Auswerteeinheit.
• 2 zeigt die Auswertung von ersten Großen. Damit wird ein Einbrechen in der Flankenzahl erkannt.
• 3 zeigt einen zeitlichen Verlauf von detektierten Flanken. Daran wird veranschaulicht, wie eine Detektion eines periodischen Einbrechens der Flankenzahl erfolgt.
• 4 zeigt die prinzipielle Funktionsweise der Zahnausfalluberwachung.
• 5 zeigt in einem Flussdiagramm den groben Ablauf des Verfahrens.
• 6 zeigt den groben Aufbau der Vorrichtung zur Überwachung von Drehzahlfuhlern auf fehlende Zähne.

An exemplary embodiment of the invention is shown and explained in the following drawing. The drawing consists of the 1 to 6 ,

• 1 shows the basic structure of a speed sensor and the associated evaluation unit.
• 2 shows the evaluation of the first big ones. A break in the number of edges is thus recognized.
• 3 shows a time course of detected edges. This is used to illustrate how a periodic drop in the number of edges is detected.
• 4 shows the basic functioning of tooth loss monitoring.
• 5 shows the flow of the procedure in a flowchart.
• 6 shows the rough structure of the device for monitoring speed sensors for missing teeth.

Exemplary embodiments

In 1 the basic structure of a speed sensor and the associated evaluation unit is shown. There the output signal (sensor signal) S1 a rotation speed sensor, consisting of the components 10 . 11 and 12 evaluated. A sensor detects this speed of rotation 10 for example the speed of rotation of a wheel by a disc rotating at wheel speed 11 with a number of teeth 12 is scanned.
The disc 11 with teeth 12 is also referred to below as a magnet wheel or an impulse wheel. The physical principle of the rotational speed sensor can be based, for example, on the fact that a voltage induced by a time-varying magnetic flux is measured. A sensor based on the principle of the Hall effect is also conceivable.

In the evaluation unit 13 , in which the method according to the invention is carried out, (by the sensor 10 supplied) sensor signal S1 evaluated. Based on the evaluation of the sensor signal S1 are in the evaluation unit 13 control signals S2 , formed for example for a wheel slip control system and / or wheel deceleration control system. This can be designed, for example, as an ABS system, an ASR system or as a vehicle dynamics control system (FDR = "vehicle dynamics control system", ESP = "Electronic Stability Program"). In addition, the evaluation unit 13 possibly other signals coming from outside S3 receive. In 1 below is an example of the temporal course of the sensor signal or the pulse sequence S1 located. 14a and 14b represent a rising or falling pulse edge. In the following, no further distinction is made between rising and falling pulse edges, and only pulse edges are mentioned. With 15 a dashed line shows a missing pulse, for example, the cause could be a missing tooth of the pulse wheel. This missing pulse leads to the absence of two edges when evaluating the pulse sequence S1 ,

$$\text{OG}=\text{obere Grenze und}$$

$$\text{UG}=\text{untere Grenze}.$$

OTOL und UTOL sind die Toleranzwerte, welche als maximale Abweichungen der Große NF/Δt1 vom Mittelwert erlaubt sind. OTOL = OG-MW ist der Toleranzwert nach oben, UTOL = MW-UG ist der Toleranzwert nach unten. Für den Wert Δt1 hat sich ein Wert von 20 Millisekunden als geeignet erwiesen (selbstverständlich sind auch andere Werte für Δt1 denkbar). Die Überwachung überprüft deshalb in jedem 20-Millisekunden-Zyklus die Anzahl der Impulsflanken an allen 4 Radern auf unerlaubtes Absinken bzw. Einbrechen. Unter den Begriffen „Absinken“ bzw. „Einbrechen“ wird dabei verstanden, dass die Große NF/Δt1 die untere Grenze UG unterschreitet. Für das Detektieren von Einbrüchen in der Flankenzahl eines 20-Millisekunden-Zyklus wird ein Band um das „edgecycle“-Signal gelegt, dessen Grenzen aus dem Mittelwert der Werte NF/Δt1 der beiden letzten Zyklen und einem Toleranzwert für die obere (OTOL) und untere (UTOL) Grenze gebildet werden. Ein Flankenzahleinbruch liegt vor, wenn die untere Grenze des Bandes verletzt wird. Ursache kann ein fehlender Zahn des Impulsrades sein, d.h. die Anzahl der erfasssten Impuls NF im Zeitintervall Δt1 wird kleiner. Die obere Grenze dient zur Erkennung positiver Gradienten der Große NF/Δt1, wie sie bei Fahrt auf Schlechtwegstrecken vorkommen können. Eine mögliche Ursache dafür sind Schuttelbewegungen des Fahrzeugs aufgrund des Schlechtwegs. Unter einem positiven Gradienten wird dabei verstanden, dass die Große NF/Δt1 die obere Grenze OG uberschreitet, d.h. es werden zuviele Flanken erfasst. Tritt in einem Zyklus Δt1 eine Grenzverletzung auf, dann wird der in diesem Zyklus ermittelte Wert NF/Δt1 nicht für die Berechnung der Grenzen im nächsten oder übernächsten Zyklus verwendet.In 2 the evaluation of the first sizes is shown. The time axis t is shown to the right.
There are discrete points in time along the time axis t K2 . tK1 and tK are shown (tK2 <tK1 <tK). The large NF / Δt1 is plotted in the ordinate direction. NF means the number of the speed sensor paid edges during a given time interval .DELTA.t1 , The large NF / Δt1 is also referred to below as the “edgecycle” and clearly means the number of edges per time interval. Furthermore, this variable is related to a wheel or a speed sensor, ie this variable is determined and evaluated for each wheel or each speed sensor. The values are still along the ordinate MW . OG and UG located. Mean $$\text{MW}=\text{Average}.$$

$$\text{OG}=\text{upper limit and}$$

$$\text{UG}=\text{lower limit},$$

OTOL and UTOL are the tolerance values that are allowed as maximum deviations of the large NF / Δt1 from the mean. OTOL = OG-MW is the upper tolerance value, UTOL = MW-UG is the lower tolerance value. For the value .DELTA.t1 a value of 20 milliseconds has proven to be suitable (other values for .DELTA.t1 conceivable). The monitoring therefore checks the number of pulse edges on all 4 wheels for unauthorized dropping or breaking in every 20 millisecond cycle. The terms “sinking” and “breaking in” mean that the large NF / Δt1 is the lower limit UG below. To detect dips in the number of edges of a 20 millisecond cycle, a band is placed around the “edgecycle” signal, the limits of which are made up of the mean value NF / Δt1 of the last two cycles and a tolerance value for the upper (OTOL) and lower (UTOL) limit. A flank drop occurs when the lower limit of the band is violated. The cause can be a missing tooth of the impulse wheel, ie the number of the detected impulse NF in the time interval .DELTA.t1 gets smaller. The upper limit is used to detect positive gradients of the large NF / Δt1, which can occur when driving on rough roads. One possible reason for this is the vehicle's shaking movements due to the poor road. A positive gradient is understood to mean that the large NF / Δt1 is the upper limit OG exceeded, ie too many edges are detected. Occurs in a cycle .DELTA.t1 If a limit violation occurs, the value NF / Δt1 determined in this cycle is not used for calculating the limits in the next cycle or the cycle after next.

The calculation of the values NF / Δt1 and the monitoring of the limits is carried out separately for each vehicle wheel or each speed sensor. If a drop in the large NF / Δt1 is detected on a wheel, ie a drop below the lower limit UG , then the monitoring is fixed on this wheel. (As already mentioned) this fixing is not absolutely necessary, but it does save resources.

For this wheel it is now determined whether the break-in, i.e. falling below the lower limit, repeated periodically. To do this, the "edgecycles" are added up using an edge counter. If the number of edges in a 20-millisecond cycle is detected again after one wheel revolution, an error counter is incremented and the accumulation of the edges begins again. This means that the number of edges per cycle (i.e. the large NF / Δt1), which is repeated cyclically with one wheel revolution, only increments the error counter.

The method for determining whether this collapse of the large NF / Δt1 is repeated periodically is shown in 3 shown.

1. 1) In Ordinatenrichtung durchgehende Striche: Jeder dieser durchgehenden Striche bedeutet eine erkannte Flanke.
2. 2) In Ordinatenrichtung unterbrochene Striche: Jeder dieser unterbrochenen Striche bedeutet eine ausbleibende Flanke. Diese ausbleibende Flanke kann z.B. durch das Fehlen eines Zahnes am Impulsrad bedingt sein.

In 3 the time t is plotted to the right. Two types of vertical lines are drawn in the figure:

1. 1) Continuous lines in the ordinate direction: Each of these continuous lines means a recognized edge.
2. 2) Lines interrupted in the ordinate direction: Each of these interrupted lines means a missing edge. This missing edge can be caused, for example, by the lack of a tooth on the pulse wheel.

• Dazu sind zwei solche Zeitintervalle Δt1 („edgecycles“) als Querbalken im unteren Teil des Bildes aufgetragen. Als Große des Zeitintervalls ist dabei beispielhaft der Wert 20 Millisekunden angenommen.
  • - Bei der Auswertung der Große NF/Δt1 beim linken und damit zeitlich früheren dieser Zeitintervalle (mit dem konkreten Wert 20 ms versehen) werden zwei fehlende Flanken detektiert. Es fehlen die beiden ersten Flanken.
  • - Deshalb werden nach Abschluss dieses Zeitintervalls Δt1 durch den Flankenzähler alle Flanken der nachfolgenden Zyklen aufsummiert, bis bei der Auswertung der Große NF/Δt1 erneut ein Einbrechen der Flankenzahl festgestellt wird. Dies ist beim rechten der beiden als Querbalken eingezeichneten Zeitintervalle Δt1 (auch hier mit dem konkreten Wert 20 ms versehen) der Fall.

As a concrete example, consider an impulse wheel with 48 teeth (the number 48 does not of course limit the generality of the invention). With a complete revolution of such an intact impulse wheel, the speed sensor counts 2 * 48 = 96 edges. In 3 the case of a pulse wheel is shown, which in the intact case 48 Teeth has, but has only 47 teeth due to a defect. Therefore, only 94 edges are paid during one revolution, this is in 3 shown with the upper crossbar. It can also be seen from the fact that left in 3 two vertical lines interrupted in the direction of the ordinate are shown, these two interrupted lines turn in on the right 3 after a full rotation of the magnet wheel again. In addition to monitoring the large NF / Δt1, an edge counter is now activated separately, which pays the edges detected by the transducer.
This is also in 3 to see:

• There are two such time intervals .DELTA.t1 ("Edgecycles") as crossbar in the lower part of the picture. The value of the time interval is an example 20 Accepted milliseconds.
  • - When evaluating the large NF / Δt1 at the left and thus earlier of these time intervals (provided with the concrete value 20 ms), two missing edges are detected. The first two flanks are missing.
  • - That is why after this time interval has ended .DELTA.t1 all edges of the subsequent cycles are summed up by the edge counter until a further drop in the number of edges is detected when evaluating the large NF / Δt1. This is the time interval shown on the right of the two as crossbars .DELTA.t1 (here also with the specific value 20 ms) the case.

• - lediglich eine Information darüber, dass wahrend der 20 Millisekunden (bzw. wahrend des Zeitintervalls Δt1) zuwenig Flanken gezahlt wurden,
• - allerdings keine Information über die zeitliche Lage der fehlenden Flanken innerhalb dieses Zeitintervalls.

The problem now arises in the evaluation that the edge counter did not count 94 edges (as it should be), but only 90 edges. This is due to the fact that the counting process of the edge counter only began after the left time interval of the length 20 ms. This means that the 4 edges still recorded in this time interval were not paid for, 94 - 4 = 90 edges were paid. The "edgecycle" signal delivers

• only information that too few edges were paid during the 20 milliseconds (or during the time interval Δt1),
• - However, no information about the temporal position of the missing edges within this time interval.

1. 1) Das Vorliegen einer vollen Radumdrehung und damit das periodische Auftreten der fehlenden Flanken wird dann angenommen, wenn der Wert des Flankenzahlers innerhalb eines vorgegebenen Auswertefensters liegt. Bei 48 Zahnen wurden innerhalb einer Radumdrehung 96 Flanken auftreten, so dass die Grenzen des Auswertefensters beispielsweise bei 90 Flanken als Untergrenze und 102 Flanken als Obergrenze liegen konnten. Liegt die Zahl der vom Flankenzahler zwischen zwei Einbruchen der Große NF/Δt1 gezahlten Flanken also zwischen 90 und 102, wird angenommen, dass eine volle Radumdrehung vorliegt. Damit wird eine Periodizitat des Einbrechens der Große NF/Δt1 angenommen und damit wird auf einen fehlenden Zahn geschlossen. Es ist auch möglich, erst nach mehrfacher Detektion eines Einbrechens der Große NF/Δt1 auf Periodizitat zu schließen.
2. 2) Eine weitere Verbesserung erfolgt dadurch, dass die Grenzen des Auswertefensters neben einem konstanten Anteil noch einen dynamischen Anteil beinhalten. Dieser dynamische Anteil verändert sich vorteilhafterweise mit der Fahrzeuggeschwindigkeit. Dies dient zur Detektierung einer vollen Radumdrehung unabhängig von der Lage des Flankeneinbruchs innerhalb eines 20-Millisekunden-Zyklus. Liegt der Flankeneinbruch nämlich am Anfang eines 20-Millisekunden-Zyklus, dann fehlen beim Aufsummieren der Flanken nach einer Triggerung, abhangig von der Geschwindigkeit der Fahrzeugs, eine bestimmte Anzahl von Flanken. Es besteht somit die Möglichkeit, dass ein konstanter Grenzwert unter Umstanden nicht erreicht werden kann.

Here there is a basic blurring, the remedy of which the invention provides for the following two options:

1. 1) The existence of a full wheel revolution and thus the periodic occurrence of the missing edges is assumed if the value of the edge number lies within a predetermined evaluation window. at 48 Teeth were within one wheel revolution 96 Flanks occur, so that the limits of the evaluation window could be, for example, 90 edges as the lower limit and 102 edges as the upper limit. If the number of flanks paid by the flank counter between two dips in the large NF / Δt1 is between 90 and 102, it is assumed that there is a full wheel revolution. A periodicity of the collapse of the large NF / Δt1 is assumed and a missing tooth is thus concluded. It is also possible to deduce periodicity only after multiple detection of a drop in the large NF / Δt1.
2. 2) A further improvement is achieved in that the limits of the evaluation window contain a dynamic part in addition to a constant part. This dynamic component advantageously changes with the vehicle speed. This is used to detect a full wheel revolution regardless of the position of the flank break within a 20 millisecond cycle. If the edge break is at the beginning of a 20 millisecond cycle, then depending on the speed of the vehicle, a certain number of edges are missing when the edges are summed up after triggering. It is therefore possible that a constant limit value may not be reached under certain circumstances.

The dynamic part recorded as 2) will be explained in more detail below. The dynamic part leads to an expansion of the limits of the evaluation window. The lower limit is reduced by the value DELTA, the upper limit is increased by the value DELTA. For example, if the previous limits were 90 and 102, the dynamically expanded limits can now be found at the values (90 - DELTA) and (102 + DELTA).

The large DELTA is calculated using the following relationship, for example: $$\text{DELTA}=\left(\text{vref}*\Delta \text{t}1*\text{Total number of edges}\right)/\text{Wheel size},$$

Where vref is the vehicle reference speed, .DELTA.t1 is the width of the time window used for the calculation of the "edge cycle" signal. The total number of flanks is the number of pulse flanks recorded during an revolution with an intact magnet wheel, ie total flank number = 2 * number of teeth.

The relationship for DELTA can also be understood very clearly if you transform it: $$\begin{array}{c}\text{DELTA}=\text{vref}/\left(2*\mathrm{\pi }*\text{r}\right)*\mathrm{<figure-callout\; id="1"\; label="\Delta \; t"\; filenames="DE000010162599B4\_0010.png"\; state="\{\{state\}\}">\Delta \; </figure-callout>}\text{t}1*\text{Total number of edges}\\ =\mathrm{\omega }/\left(2*\mathrm{\pi }\right)*\mathrm{<figure-callout\; id="1"\; label="\Delta \; t"\; filenames="DE000010162599B4\_0010.png"\; state="\{\{state\}\}">\Delta \; </figure-callout>}\text{t}1*\text{Total number of edges}\\ =\Delta \mathrm{\phi }/\left(2*\mathrm{\pi }\right)*\text{Total number of edges}\end{array}$$

ω is the angular velocity of the wheel, Δφ = ω * .DELTA.t1 is the angle of rotation with respect to that in the time interval .DELTA.t1 rotation of the wheel (in other words: in the time interval .DELTA.t1 the wheel has turned the angle Δφ turned). Thus Δφ / (2 * π) * total number of edges is the number in the time interval .DELTA.t1 determined edges for a wheel at the current vehicle speed. The vehicle speed is great Δφ contain.

• - Hat der Fehlerzahler den Wert Null erreicht, wird die Überwachung für die anderen Rader wieder freigegeben (sofern die anderen Rader überhaupt von der Überwachung ausgenommen wurden).
• - Überschreitet der Fehlerzahler seinen Grenzwert, erfolgt ein Fehlereintrag in einen Fehlerspeicher. Anschließend wird die Überwachung zurückgesetzt und kann an den restlichen drei Rädern fortgesetzt werden.

The error payer is only decremented if, after two consecutive wheel revolutions, the number of edges within the Evaluation window was recognized. As soon as an edge break is again recognized within the evaluation window and the error payer has a value> 0, the error payer is incremented again.

• - If the number of errors has reached zero, the monitoring for the other bikes is released again (if the other bikes were excluded from the bumping at all).
• - If the error payer exceeds its limit value, an error entry is made in an error memory. The monitoring is then reset and can be continued on the remaining three wheels.

$$\text{VR}=\text{rechtes Vorderrad},$$

$$\text{HL}=\text{linkes Hinterrad und}$$

$$\text{HR}=\text{rechtes Hinterrad}.$$

The basic functioning of tooth loss monitoring is based on 4 explained. From top to bottom are the signal curves of the sizes NF / Δt1 for the four vehicle wheels VL . VR . HL and MR applied. To the right is time t applied. Mean $$\text{VL}=\text{left front wheel}.$$

$$\text{VR}=\text{right front wheel}.$$

$$\text{HL}=\text{left rear wheel and}$$

$$\text{MR}=\text{right rear wheel},$$

• - Zum Zeitpunkt t1 wird ein Einbrechen des Wertes NF/Δt1 für das linke Vorderrad (in der obersten Reihe mit VL bezeichnet) festgestellt.
• - Als Folge wird nun die Überwachung auf dieses Rad konzentriert, die Überwachung der übrigen Rader wird vorübergehend unterbrochen (wie bereits erwähnt nicht zwangsläufig, auch eine weitere Überwachung der übrigen Rader ist moglich). Dies ist durch die fehlende leichte Grauschraffur für die übrigen Rader (VR, HL, HR) für t > t1 zu erkennen.
• - Nun muss untersucht werden, ob sich dieses Einbrechen von NF/Δt1 nach einer Radumdrehung wiederholt. Dabei wird nun die Flankenzahl bis zu einem erneuten Einbrechen ermittelt, wie es bereits beschrieben wurde.
• - Es wird ein erneuter Einbruch detektiert, die hochgezahlte Flankenzahl liegt auch innerhalb des zulassigen Auswertefensters. In 4 sind die Zahlen 90 und 102 (welche beispielhaften Charakter haben) als Grenzwerte für die Flankenzahl, d.h. als Grenzwerte des Auswertefensters eingezeichnet.
• - Da nun angenommen werden darf, dass dieselbe fehlende Zahnlücke im Polrad vorliegt, wird der Fehlerzähler um den Wert 1 inkrementiert.
• - Dieser Vorgang wird nun immer wieder wiederholt, dies ist in der obersten Zeile (VL) mit den Punkten „.....“ gekennzeichnet.
• - Zur Zeit t2 wird zum wiederholten Mal das Einbrechen der Größe NF/Δt1 detektiert. Erneut wird der Fehlerzahler inkrementiert und er habe nun einen einen vorgegebenen Schwellwert erreicht. Deshalb erfolgt nun ein Fehlereintrag für das linke Vorderrad (VL), die Überwachung des diesem Rad zugeordneten Drehzahlfuhlers wird im folgenden ausgesetzt (erkennbar durch die nun fehlende leichte Grauschraffur für das Rad VL), die Überwachung für die übrigen drei Drehzahlfuhler wird wieder aufgenommen (erkennbar durch die nun wieder einsetzende leichte Grauschraffur für die drei Rader VR, HL und HR).

Understanding image 4 is easiest if you follow the following points:

• - At the time t1 the value of NF / Δt1 for the left front wheel (in the top row with VL designated) determined.
• - As a result, the monitoring is now concentrated on this wheel, the monitoring of the other wheels is temporarily interrupted (as already mentioned, not necessarily, further monitoring of the other wheels is also possible). This is due to the lack of light gray hatching for the other wheels ( VR . HL . MR ) for t> t1.
• - Now it has to be examined whether this collapse of NF / Δt1 repeats after one wheel revolution. The number of edges until a new break-in is now determined, as has already been described.
• - A further break-in is detected, the high number of edges also lies within the permissible evaluation window. In 4 are the numbers 90 and 102 (which have an exemplary character) as limit values for the number of edges, ie as limit values of the evaluation window.
• - Since it can now be assumed that there is the same missing tooth gap in the magnet wheel, the error counter is incremented by the value 1.
• - This process is now repeated again and again, this is in the top line ( VL ) marked with the points ".....".
• - At time t2, the collapse of the size NF / Δt1 is detected again. The error payer is incremented again and it has now reached a predetermined threshold value. Therefore, an error entry is now made for the left front wheel ( VL ), the monitoring of the speed sensor assigned to this wheel is suspended in the following (recognizable by the lack of light gray hatching for the wheel VL ), the monitoring for the remaining three speed sensors is resumed (recognizable by the light gray hatching for the three wheels, which has now started again VR , HL and HR).

In addition, bad road detection is implemented in the monitoring. With this rough road detection, all values of the large NF / Δt1 are outside the permissible limits UG and OG and added up outside the evaluation window (ie no periodicity is determined) by means of a bad road payer. If the bad road payer reaches too high a payer level, the monitoring is reset and reinitialized.

The rough procedure for monitoring speed sensors for missing teeth is in 5 shown.
In block 100 the output signal is read. Then in block 101 determined a first size, namely the size NF / Δt1. In block 102 it is checked whether this first variable falls below a lower limit value UG. If not, it will block 100 the reading of the output signal of the transducer continues. However, falls below the first size in block 102 the lower limit UG , then in block 103 found that there is a suspicion of error and the monitoring is limited to the affected speed sensor. In block 104 it is checked whether the value falls below the limit UG after each full wheel revolution, ie whether there is periodicity. If there is no periodicity, then it becomes a block 100 zuruckverzweigt. Is periodicity of the kind that falling short of UG occurs several times after a wheel revolution, before, then in block 105 the speed sensor in question is excluded from monitoring and the wheel is marked as faulty.

The rough structure of the device for monitoring speed sensors for missing teeth is in 6 shown.

The speed sensor 200 provide input signals S1 to the investigative means 201 , In the investigative means 201 first large NF / Δt1 are determined. The output signals of these determination means go to comparison means 202 , in which the values of the first large values are compared with lower limit values. The output signals of the comparison means become, depending on the result of the comparisons carried out there, the determination means 201 fed and / or they are sent to error handling agents 203 forwarded, in which it is determined whether a defective speed sensor is present. For this determination, there is a return of output signals from the error handling means to the determining means 201 necessary.

With the described method, several wheels with faulty pulse wheels can also be recognized, i.e. surveillance is not limited to just one wheel. By evaluating the number of edges instead of wheel speeds, the filter effects of speed processing can be avoided. At higher speeds, these lead to a reduction in the drops in the wheel speed signal and thus make it more difficult to identify the missing teeth.

In the exemplary embodiment, the value was frequently for the time interval Δt1 20 Milliseconds used. It goes without saying that other time periods can also be considered. In particular, the time intervals do not always have to be of the same length. Rather, time intervals of different lengths are also conceivable (provided the actual length is known in each case). An embodiment of the invention can for example look such that the time intervals Δt1 become shorter with increasing vehicle speed.

Claims (4)

Method for monitoring at least one speed sensor, which emits an output signal (S1) consisting of a sequence of pulses, whereby - at predeterminable times, a first variable (NF / Δt1) assigned to the speed sensor is determined, which is related to the number of signals in one the time interval (Δt1) correlated with the respective point in time (NF), the first variable (NF / Δt1) is compared with a calculated lower limit value (UG), with at least the first variable (NF / Δt1), which was determined at a previous point in time, and - to monitor the speed sensor, depending on the result of this comparison, a suspected error signal is generated, characterized in that a first suspected error signal is generated as soon as the value falls below the lower limit in at least one comparison Limit value (UG) by the first variable (NF / Δt1) assigned to the speed sensor f It is determined, after the first suspected error signal has been issued for those speed sensors for which a first suspected error signal has been generated, it is determined whether the drop below the respective lower limit value is repeated at a later time, wherein - the speed sensor consists of a pulse wheel (11, 12) and a transducer (10), - and the later point in time is characterized in that the impulse wheel has made one full revolution and - that the full revolution of the impulse wheel is determined by the fact that between the detection of an initial drop below the lower limit value (UG) and the detection of a second drop below the lower limit value (UG) determines a second variable correlated with the number of pulses that occur, which must lie in a predeterminable interval for the detection of a full revolution, the width of the predeterminable interval being dependent on the current vehicle reference speed (v ref) is dependent. Procedure according to Claim 1 , characterized in that after the generation of the first suspected error signal, the monitoring is limited to those speed sensors whose assigned first quantities have been determined to have fallen below the respective lower limit value (UG), Procedure according to Claim 1 , characterized in that the second variable correlated with the number of pulses occurring is the number of pulse edges. Device for monitoring at least one speed sensor (200), which emits an output signal consisting of a sequence of pulses, - in which a first variable (NF / Δt1) assigned to the speed sensors is determined in the determining means (201), which can be determined using the number of pulses (NF) occurring in a time interval correlated with the respective time point is correlated, whereby - in comparison means (202) the first variable (NF / Δt1) is compared with a calculated lower limit value (UG), in which the Calculation of the lower limit value (UG) receives at least the first variable that was determined at a previous point in time and - for monitoring the speed sensor (200) depending on the result of these comparisons A suspected error signal is generated, characterized in that a first suspected error signal is generated as soon as, in at least one comparison, a drop below the lower limit value (UG) is determined by the first variable (NF / Δt1) assigned to the speed sensor, after the first suspected error signal has been given for those Speed sensor, for which a first suspected error signal has been generated, is determined whether the drop below the respective lower limit value is repeated at a later time, wherein - the speed sensor consists of a pulse wheel (11, 12) and a sensor (10), - and the later point in time is characterized by the fact that the impulse wheel has made a full revolution and - that the full revolution of the impulse wheel is determined by the fact that between the detection of an initial drop below the lower limit value (UG) and the detection of a second drop below the lower limit value ( Basement) e second quantity correlated with the number of pulses occurring is determined, which must lie in a predeterminable interval in order to detect a full revolution, the width of the predeterminable interval being dependent on the instantaneous vehicle reference speed (vref).

Images