GB2379022A

GB2379022A - Method of error testing a sensor

Info

Publication number: GB2379022A
Application number: GB0214892A
Authority: GB
Inventors: Peter Wirth
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2001-06-30
Filing date: 2002-06-27
Publication date: 2003-02-26
Anticipated expiration: 2022-06-27
Also published as: GB2379022B; GB0214892D0; JP2003043133A; DE10131760A1; DE10131760B4

Abstract

A method of testing a sensor to detect faulty operation is carried out by intentional disturbance of the detection magnitude to be measured by the sensor and evaluation of the resulting sensor output, BB. The method comprises ascertaining, in a learning phase, the effects of the disturbance on the signal delivered by the sensor, and processing the signal BB so that the change in the sensor signal caused by the disturbance is eliminated. The processing is carried out during the testing of the sensor, and provides a cleaned signal AA representing the magnitude to be measured. The cleaned signal AA is available during continuous testing of the sensor. Sensor error may be determined from the disturbed sensor signal before processing BB, or after processing AA. Error determination from the cleaned signal may involve differentiation with respect to time.

Description

METHOD OF TESTING A SENSOR

The present invention relates to a method of testing a sensor, especially a method of testing a sensor by an intentional disturbance of the detection of the magnitude to be measured and evaluation of the resulting sensor output signal.

Such a method is used, for example, in order to test the orderly functioning of a rate-of-

rotation sensor. Rate-of-rotation sensors serve for detecting the rotational movement of an object per unit of time (as) and are used, for example, in motor vehicles in order to recognise incipient loss of control perhaps in the case of travel with excessive speed around a bend (break-away or entry into a slide) so that suitable countermeasures -

usualiy automatic braking of individual wheels - can be initiated.

This is a very important aspect of safety. Accordingly, and because braking of individual wheels, for example braking during straight-line departure, without actual cause can have disastrous effects, it is important that the sensor always functions in orderly manner and in a given case is switched off or ignored.

Such rate-of-rotation sensors are therefore constantly tested during operation. This is carried out in the manner that the actual magnitude to be measured is intentionally disturbed and the sensor output signal resulting therefrom of the sensor is evaluated.

A known procedure is illustrated in Fig. 1, which shows two different output signals of a rate-of-rotation sensor and, in particular, on the one hand a signal delivered by the sensor when measuring has not been disturbed (curve A marked by squares) and on the other hand a signal delivered by the sensor when measuring is disturbed for test purposes (curve B marked by triangles); the signal represented by curve A is termed, in the following' actual signal and the signal represented by curve B is termed measured signal.

In the illustrated example, the measurement of all 42 time units (for example, all 21 ms) is disturbed. The disturbances have the effect that the actual signal to be measured by the sensor (curve A) is superimposed by an offset, wherein this offset causes deflection of the sensor signal alternately upwardly (disturbance at t = 0) and downwardly (disturbance at t =42).

Checking of the signal is carried out in the manner that at a specific instant, for example at t = 24 and at t = 66 (generally: at T = n * 42 + 24; n = 0, 1, 2,...), it is checked whether the difference between the sensor signal delivered at the relevant instant and the last ascertained actual travel signal lie within a specific value range (rate-of-rotation range).

If that is the case, it can be concluded therefrom that the sensor operates in orderly manner; if that is not the case, it has to be concluded that the sensor does not operate in error-free manner Through such an investigation it is also possible to check the sensor during normal use thereof. However, it is disadvantageous that the sensor can then only be used for detection of the magnitude when the sig..ai B it has Just delivered has not been influenced by a disturbance, thus corresponds to the actual travel signal A. In the example under consideration this is possible at, for example, t = 40 and t = 82 and again at intervals of 42 time units.

The intervals in time at which the sensor delivers the actual travel signal are, in consideration of the fact that there has to be immediate reaction to risk situations that may arise, relatively large. On the other hand, checking of the orderly functioning of the sensor should never be interrupted for a longer time, whereby there is no apparent alternative to the described procedure.

There is therefore a need for a method by which, even when the sensor is tested without interruption or without greater interruption, a signal representing the magnitude to be measured is available at very short intervals in time.

According to the present invention there is provided a method of testing a sensor by an intentional disturbance of the detection of the magnitude to be measured and evaluation of the output signal, which results therefrom, of the sensor, characterized in that in a learning phase the effects of the disturbance on the signal delivered by the sensor are ascertained and that during the testing of the sensor the signal delivered by this is so processed that the change, which is caused by the disturbance, in the sensor output signal is eliminated.

Preferably, the detection of the magnitude to be measured is disturbed in such a manner that the effects, which result therefrom, on the output signal of the sensor are independent of the amount and the course of the magnitude to be measured.

For preference, in the learning phase a quantitative ascertaining of the effects of the disturbance on the sensor signal is carried out and/or the course over time of such effects is carried out. Preferably, the effects, which are ascertained in the learning phase, of the disturbances on the signal delivered by the sensor are stored.

The ascertaining of the effects of the disturbances on the signal delivered by the sensor can be carried out in a phase in which it is known which signal of the sensor was delivered without a disturbance of the ascertaining of the magnitude to be measured. In addition, the output signal, which is freed from the effects caused by the disturbance, of the sensor can be subjected to a specific processing and the signal, which is obtained as a result of this processing, is thereupon investigated with respect to whether it fulfils specific conditions. A method exemplifying the invention is distinguished by the fact that in a learning phase the effects of the disturbance on the signal delivered by the sensor are ascertained and that during the testing of the sensor the signal delivered by this is processed so that the change, which is caused by the disturbance, in the sensor output signal is eliminated. As a result, even in phases in which the signal delivered by the sensor is influenced by the disturbances carried out for testing the sensor the signal which the sensor would deliver without these disturbances can be determined.

An example of the method according to the invention will now be more particularly described with reference to the accompanying drawings, in which: Fig. 1 is, as already described, a diagram showing the signal which a rate-of rotation sensor delivers when it is tested and the signal which the sensor would deliver if it has not been tested; Fig. 2 is a diagram similar to Fig. 1 showing the signal which a rate-of-rotation sensor delivers when it is tested, a cleaned signal ascertained therefrom by a method exemplifying the invention and a further signal obtained from a differentiation of the cleaned signal; and

Fig. 3 is a block circuit diagram of means for carrying out the method.

Referring now to the drawings, the sensor, on the basis of which the method described in more detail in the following is explained, is again a rate-of-rotation sensor, but it could in principle be any other kind of sensor.

This sensor is tested, as in the case of the conventional method, continuously during the normal operation thereof. The test is again carried out by a disturbance, which causes generation of an offset in the sensor signal, in the detection of the actual magnitude to be measured and evaluation of the resulting output signal of the sensor.

The resulting output signal of the sensor is illustrated in Fig. 2 by the curve BB.

The curve BB shown in Fig. 2 results from the fact that the detection of the actual magnitude to be measured is so disturbed at the interval of 42 time units that the actually delivered sensor signal is superimposed by an offset The superimposed offset is the same offset as in the curve B according to Fig. 1. That the curve BB nevertheless has a different path from that of the curve B lies in the fact that the actual travel signals differ.

The actual travel signal on which the curve BB is based is illustrated by a curve AA in Fig 2, wherein the curves M and A have recognizably different paths If the actual magnitudes to be measured, i e the curves A and M were the same, the curves B and BB would have the same path. To that extent, the conventional method described in the introduction and the new method correspond with one another

There is no restriction on the fact that the curves B and BB have the same path in the case of identical starting conditions. In particular, in the case of the method exemplifying the invention an offset causing any other course of the curve BB can be used. Only relatively minor conditions are imposed on the offset. It only has to be guaranteed by the offset employed that the changes which the sensor output signal thereby experience are qualitatively and quantitatively the same each time and not dependent on the amount and the course of the magnitude to be measured. The changes are preferably, not dependent on the prevailing temperature and the age of the sensor, although compensation could be made for temperature influences and also ageing influences.

As in the case of the conventional method described in the introduction, based on the

signal changed by the offset (curve B or BB) it is ascertained whether the sensor operates in orderly manner. However, this takes place in the method exemplifying the invention in a manner different from the conventional method described in the introduction.

In the case of the method exemplifying the invention, there is subtracted from the signal BB delivered by the sensor the respective superimposed offset and it is checked whether the cleaned signal resulting therefrom can be a signal representing the magnitude to be measured. The offset subtracted from the sensor signal M is determined and stored in a learning phase. The learning phase is carried out in a phase in which the time course of the actual magnitude to be detected is known. This phase is preferably a phase in which the sensor, without loading with an offset, has to deliver a known constant value, this value preferably being zero. In the case of a rate-of-rotation sensor this is, for example, a phase in which it or the system in which it is incorporated is static. In such a state the detection of the actual magnitude to be detected is disturbed in the mode and manner also practiced later in normal operation of the sensor and the deviation of the sensor signal then delivered from the signal which the sensor would have delivered if it had not been disturbed is quantitatively ascertained. in the present example the difference between these signals is determined in terms of sign and amount. This determination takes place at one or more specific instants after activation of the offset, in the present example every two time units, thus after each ms. The thus- ascertained deviations are stored.

It should be noted that the ascertained deviations do not necessarily have to be the stated difference. It only has to be ensured that the actual travel signal, i.e. the signal which the sensor would have delivered if the measurement had been undertaken without offset, can be ascertained from the signal delivered by the sensor and the ascertained deviations.

Moreover, there is no absolute necessity to determine the deviations at the stated instants.

In principle, the deviations can be ascertained at however many and at whatever fixed instants.

If, as in the present example, alternately different disturbances producing different offsets are carried out, the deviations are ascertained for each of the different disturbances.

The ascertained and stored deviations are employed in order, when the signal is disturbed in normal operation for test purposes, to ascertain the signal which the sensor would have delivered if it had not been disturbed. This happens in the present example in the manner that the deviations ascertained in the test phase are subtracted from or added to the signal delivered by the sensor. For that purpose, the actual values of the signal delivered by the sensor are ascertained at those instants for which the deviations are known and the respectively associated deviations are subtracted from (or added to) these values.

The result of this correction is a cleaned signal which the sensor would have delivered if the measurement had been undertaken without disturbance, thus the actual travel signal i!!ustrated by the curve M in Fig. 2. Thus, in the case o, the described method, signals representing the magnitude to be measured are available even during testing of the sensor and independently of the deviations ascertained in the test phase - however many successive signals at whatever intervals in time.

Based on the cleaned signal M it can also be tested whether the sensor operates in orderly manner. For this purpose, the cleaned signal M is differentiated according to time and it is checked whether the result of this differentiation exceeds a predetermined positive or a predetermined negative limit value. The result of the mentioned differentiation represents the rate-of-rotation change per unit time and is illustrated in Fig. 2 by the curve denoted by the reference symbols CC.

If, and for as long as, the signal CC does not exceed the mentioned limit values, it can be inferred therefrom that the sensor operates in orderly manner; if the signal CC exceeds one of the limit values, this is an indication of the fact that the sensor does not operate in orderly manner. This is based on recognition that the object, the rate of rotation of which is to be ascertained by the sensor, can change its rate of rotation per unit time only to a specific degree, this degree generally being smaller the larger and/or heavier the object is.

For example' a motor vehicle can change its rate of rotation, even in exceptional situations, at most by a specific extent per unit time. If, however, the curve CC exhibits a change in rate of rotation which is greater than the maximum value ascertained by tests or estimated, then it can be concluded therefrom that the sensor delivers an incorrect value.

It would also be conceivable to subject the signal CC to a different processing, for example an integration, and to make the determination of whether or not the sensor operates in orderly manner dependent on the result of this processing.

Whether the sensor operates in orderly manner can, however, also be ascertained as before directly from the signal BB delivered by the sensor, for example by ascertaining whether the instantaneous amplitude of this signal at predetermined instants is disposed within a specific range. It would also be conceivable to process the signal BB only in a specific way, for example by differentiation or integration, and to check whether the signal or result resulting therefrom fulfils specific conditions.

Different investigations can also be carried out simultaneously, successively or alternatively. Depending on the mode and manner in which it is checked whether or not the sensor operates in orderly manner, values representing the signal (curve M) to be measured are, however, always available Fig. 3 shows means by which the afore-described method can be realised, this means comprising a sensor 1, a sensor signal receiving device 2, a test control device 3, a deviation ascertaining device 4, a deviation storage device 5, an error recognition device 6, a selector device 7, a subtracter 8 and a clock generator 9.

The sensor 1 is the sensor to be tested. It detects the measurement magnitude to be ascertained and delivers an analog signal corresponding thereto and in addition includes means which, at the instigation of the test control device 3, produce a disturbance in the detection of the magnitude to be measured (for the offset generation).

The signal receiving device 2 receives the signal delivered by the sensor 1 and conducts it to the subtracter 8, the deviation ascertaining device 4 and the error recognition device 6.

The deviation ascertaining device 4 ascertains, in the learning phase, the deviation between the sensor output signal fed by the signal receiving device 2 and the sensor output signal which would have delivered in the prevailing conditions without a disturbance of the detection of the magnitude to be measured. Data representing the ascertained result (the ascertained deviation) are delivered to the deviation storage device 5.

The deviation storage device 5 stores the data fed thereto by the deviation ascertaining device 4 and delivers data as required to the error recognition device 6 and to the selector device 7.

The selector device 7 receives the output signal of the deviation storage device 5 and a signal having the value O and establishes which of these signals is fed to the subtracter 8.

The subtracter 8 subtracts the signal, which is fed from the selector device 7, from the signal fed thereto by the sensor signal receiving device 2. The output signal is the signal representing the amount of the magnitude to be measured by the sensor 1 (the cleaned signal or the actual travel signal; curve M).

The error recognition device 6 receives the output signals of the sensor signal,eceiving device 2 and the deviation storage device 5, forms the difference between these signals, differentiates the result according to time and checks whether the result of the differentiation lies within permissible limits. If this is not the case, a signal is delivered to indicate that the sensor does not operate in orderly manner.

The test control device 3 controls the deviation ascertaining device 4, the deviation storage device 5, the error recognition device 6, the selector device 7 and the clock generator 9 and causes as already noted, the disturbance required for testing of the sensor.

Both the ascertaining which is undertaken in the learning phase, of the effect of the disturbance on the signal delivered by the sensor 1 and the normal operation of the system can thereby be controlled by the test control device 3.

In the learning phase the test control device 3 causes generation of the disturbance, which is also used in normal operation for testing the sensor, and ensures that the deviation ascertaining device 4 ascertains the effect of the disturbance on the sensor signal and that the data representing the ascertained effect are stored in the deviation storage device 5.

Through activation and deactivation of the clock generator connected with the deviation ascertaining device 4 and the deviation storage device 5 the test control device 3 additionally predetermines at which instants (referred to the start of the disturbance) the effect is to be ascertained and stored.

In normal operation, the test control device 3 ensures that disturbances are generated at short intervals in time, that the data delivered by the deviation storage device 5 always contain the deviations ascertained in the test phase at the respective instants and that the selector device 7 passes on to the subtracter the signal fed thereto by the deviation storage device 5.

The described method and means for performing the method make it possible to provide at desired short intervals in time a signal representing the magnitude to be measured independently of the specific details of the practical realization, even when the sensor is tested without, or without greater, interruption.

Claims

1. A method of testing a sensor for detecting a magnitude, comprising the steps of intentionally disturbing detection of the magnitude by the sensor, evaluating an output signal of the sensor provided as a result of the disturbance so as to recognise faulty functioning of the sensor, ascertaining in a learning phase the effect of the disturbance on the signal and processing the signal in dependence on the ascertained effect to eliminate change in the signal caused by the disturbance.

2. A method as claimed in claim 1, wherein the step of disturbing is carried out in such a manner that the effect of the disturbance on the signal is independent of the amount and course of the magnitude to be detected.

3. A method as claimed in claim 1 or claim 2, wherein the step of ascertaining is carried out quantitatively.

4. A method as claimed in any one of the preceding claims, wherein the step of ascertaining is carried out to determine the course over time of the effect of the disturbance on the signal.

S. A method as claimed in any one of the preceding claims, comprising the step of storing the ascertained effect.

6. A method as claimed in any one of the preceding claims, wherein the step of ascertaining is carried out in a phase in which it is known that the sensor without disturbance of the detection of the magnitude delivers a predetermined output signal.

7. A method as claimed in any one of the preceding claims, comprising the step of subjecting the signal from which the change has been eliminated to a specific form of processing and thereafter checking the signal for fulfillment of a predetermined condition associated with the specific processing.