DE102014200426A1 - Method for robust detection of high-frequency signal interference - Google Patents

Abstract

The invention relates to a method and a corresponding device for detecting signal disturbances in a sensor signal, in particular in a sensor signal of a vehicle sensor. A method for detecting a signal disturbance of a sensor signal is described. The sensor signal comprises a sequence of measured values xk to a corresponding sequence of times k. The method comprises determining a reference signal by applying a Kalman filter to the sensor signal. The reference signal comprises a sequence of reference values x∨k. The method further comprises detecting a signal disturbance of the sensor signal based on the reference signal.

Description

The invention relates to a method and a corresponding device for detecting signal disturbances in a sensor signal, in particular in a sensor signal of a vehicle sensor.

In vehicles (eg, in a passenger car, a truck, or a motorcycle), a variety of sensors are used, including: a. to record measurement data on the driving dynamics of the vehicle. For example, vehicles include one or more yaw sensors to detect rotational movement of the vehicle about the vertical axis of the vehicle. The sensor signals detected by the one or more sensors can then be used to carry out measures to improve the driving dynamics of the vehicle. For example, a sensor signal detected by a yaw sensor may be used to provide an electronic stability program in the vehicle.

In order to be able to reliably use the sensor signal detected by a sensor to provide vehicle functions (such as providing an electronic stability program), it should be ensured that the sensor signal of a sensor reflects an actual situation (eg the sensor signal of a yaw sensor represents an actual yaw motion of the vehicle). If there are reasonable doubts about the correctness of a sensor signal, then this sensor signal should be excluded in the provision of vehicle functions. For safety reasons, for example, vehicle functions that use the sensor signal should be switched off or reduced in their functional scope.

In order to avoid unnecessary switching off of vehicle functions, on the other hand, it should be ensured that short-term signal disturbances of the sensor signal (for example due to short-term interference or other disturbances) are detected quickly and reliably as such, and thus do not lead to the deactivation of a vehicle function ,

The present document deals with the technical task to identify such short-term signal interference of a sensor signal in a reliable and timely manner, and if necessary to differentiate from measurement errors of the sensor signal.

The object is solved by the independent claims. Advantageous embodiments are u. a. in the dependent claims.

In one aspect, a method of detecting a signal disturbance of a sensor signal is described. The sensor signal comprises a sequence of measured values x _{k relative} to a corresponding sequence of times k. The sensor signal may have been detected by a sensor of a vehicle (eg, a passenger car, a truck, or a motorcycle). The sensor signal may comprise measured values x _{k of} a yaw sensor. Thus, the sensor signal may include a rotation angle and / or a yaw angle of the vehicle about a vertical axis of the vehicle.

The method comprises determining a reference signal by applying a Kalman filter to the sensor signal. In other words, the reference signal can be determined by Kalman filtering of the sensor signal. In this case, the reference signal may comprise a sequence of reference values x ∨ _k . In particular, the reference signal may include corresponding reference values for the measured values of the sensor signal. In other words, the reference signal may comprise respective reference values for the measured values of the sensor signal.

The method further comprises detecting a signal disturbance of the sensor signal based on the reference signal. In particular, it can be distinguished or disambiguated on the basis of the reference signal in a reliable and timely manner between a (short-term) signal disturbance and an (inherent) measurement error of the sensor signal. The use of a Kalman filter makes it possible to determine a reference signal which has little or possibly no phase delay to the sensor signal. This enables a reliable and timely detection of signal interference by comparing the reference signal with the sensor signal.

In particular, the method may include determining a deviation ε _k between a measured value x _k and a corresponding reference value x ∨ _k at the time k. Deviations for the sequence of times can be determined. On the basis of the determined deviation ε _k , an indication for a signal disturbance or for a measurement error of the sensor signal can be determined. In particular, an indication can be determined that there is either a signal disturbance or a measurement error of the sensor signal at the time k. In order to determine such an indication, it may be determined that an amount of the determined deviation ε _{k is} equal to or greater than a threshold value.

As stated above, the use of a Kalman filter enables the reduction of phase delay between the sensor signal and the reference signal. This in turn allows the determination of indicia based on the determined Deviations ε _k between measured values x _k and corresponding reference values x ∨ _k .

The threshold for determining an indication may depend on an energy of a noise component of the sensor signal. The noise component of the sensor signal can, for. B. be determined by a high-pass filtering of the sensor signal. With increasing energy of the noise component of the sensor signal, the threshold value can rise. By taking into account the energy of the noise component in determining an indication of a signal disturbance or for a measurement error of the sensor signal, the detection of signal disturbances can be made more robust.

The method may further include disambiguating between a signal disturbance or a measurement error of the sensor signal at time k, based on the reference signal. In particular, it can be determined on the basis of the determined deviations ε _k between measured values x _k and corresponding reference values x ∨ _k , whether the determined index at time k is due to a signal disturbance or to a measuring error.

The disambiguation may include determining a further deviation ε _{k + 1} between a measured value x _{k + 1} and a corresponding reference value x ∨ _{k + 1} at a time k + 1, which is after the time k at which the indicia for a Signal error or was determined for a measurement error of the sensor signal. In other words, deviations ε _{k + 1} at times after time k can be determined and analyzed. In particular, on the basis of the further deviation ε _{k + 1, it} can be determined that at a certain point in time k + 1 there is another indication of a signal disturbance or of a measurement error of the sensor signal. For this purpose, the amount of deviation ε _{k + 1} can be compared with the above-mentioned threshold value (which may depend on the energy of the noise component of the sensor signal).

It can then be disambiguated or distinguished between a signal disturbance or a measurement error of the sensor signal on the basis of the deviation ε _k and on the basis of the further deviation ε _{k + l} . In particular, it can be determined whether the deviation ε _{k has} a different sign than the further deviation ε _{k + l} . A different sign typically indicates the presence of a signal disturbance. Alternatively or additionally, it can be determined whether a time interval between the time k at which the indication for a signal disturbance or for a measurement error of the sensor signal was determined, and the time k + l, at which the further indication for a signal disturbance or for a measurement error of the sensor signal is equal to or less than a predefined time period. The presence of two indicia within the predefined time period (eg 50 ms or less) is an indication of the presence of a signal disturbance (especially if the deviations have a different sign).

Thus, it can be determined in an efficient and reliable manner, based on the reference signal and the sensor signal, whether the sensor signal comprises a signal disturbance or a measurement error. This can increase the robustness and safety of vehicle functions based on the sensor signal.

The Kalman filter may describe a state of the vehicle through a state vector X. The state vector X at the time k may comprise the sensor signal at the time k (ie the measured value x _k ), as well as one or more derivatives of the sensor signal at the time k. A state transition of the vehicle between states at different times may be described by a third or higher order polynomial. The use of such a state model makes it possible to determine in advance the feedback or Kalman gains in the Kalman filter. This in turn allows resource-efficient calculation of the reference signal during operation of the vehicle.

In another aspect, a control unit for a vehicle is described. The vehicle includes a sensor (eg, a yaw sensor) for detecting a sensor signal. The sensor signal comprises a sequence of measured values x _{k relative} to a corresponding sequence of times k. The control unit is set up to determine a reference signal by applying a Kalman filter to the sensor signal. The reference signal comprises a sequence of reference values x ∨ _k corresponding to the sensor signal. The control unit is further configured to detect a signal disturbance of the sensor signal on the basis of the reference signal.

The control unit may be further configured to disambiguate a signal disturbance of the sensor signal from a measurement error of the sensor signal on the basis of the reference signal (eg using the method described in this document). Furthermore, the control unit may be configured to cause a functional range of a vehicle function using the sensor signal to be reduced if a measurement error of the sensor signal is detected. Alternatively or additionally, the control unit may be configured to cause the functional range of the vehicle function, which uses the sensor signal, to be maintained if a signal disturbance of the sensor signal is detected.

Thus, the control unit may be configured to increase the robustness and safety of vehicle functions (eg, from an electronic stability program) based on a reliable and timely disambiguation of signal disturbances and measurement errors.

In another aspect, a vehicle (eg, a passenger car, a truck, or a motorcycle) is described that includes a control unit described in this document.

In another aspect, a software (SW) program is described. The SW program may be set up to be executed on a processor (eg, on a control device of a vehicle), thereby performing the method described in this document.

In another aspect, a storage medium is described. The storage medium may include a SW program that is set up to run on a processor and thereby perform the method described in this document.

It should be understood that the methods, devices and systems described herein may be used alone as well as in combination with other methods, devices and systems described in this document. Furthermore, any aspects of the methods, apparatus, and systems described herein may be combined in a variety of ways. In particular, the features of the claims can be combined in a variety of ways.

Furthermore, the invention will be described in more detail with reference to exemplary embodiments. It shows

1 a block diagram of an exemplary vehicle;

2a an exemplary sensor signal with a momentary signal disturbance;

2 B an exemplary sensor signal with a measurement error;

3 an exemplary sensor signal and an exemplary reference signal determined therefrom; and

4 a flowchart of an exemplary method for detecting a signal disturbance in a sensor signal.

As set forth above, the present document deals with the technical task of detecting signal interference in a sensor signal in a reliable and timely manner and possibly disambiguating measurement errors in the sensor signal. 2a shows a sensor signal 200 with a short-term signal fault 201 , Such a signal disturbance 201 shall be recognized in a timely and reliable manner by the procedures described in this document. Furthermore, it should be ensured that such a signal disturbance 201 in a reliable way of one in the sensor signal 200 contained measurement error 202 can be distinguished (see 2 B ).

1 shows a block diagram of an exemplary vehicle 100 , The vehicle 100 includes an exemplary sensor 102 (eg a yaw sensor). The sensor 102 captures measurement data (eg a rotational movement of the vehicle 100 around a vertical axis of the vehicle 100 ) and provides the measurement data in the form of a sensor signal 200 a control unit 101 available. The sensor signal 200 typically includes a sequence of measurement data for a corresponding sequence of times. The sensor 102 is typically set up to provide measurement data with a specific measurement frequency (eg 1 measurement per ms or per μs). The sensor signal 200 thus includes a sequence of measurement data with the appropriate measurement frequency.

The control unit 101 is set up, the sensor signal 200 evaluate. In particular, the control unit 101 set up a signal fault 201 the sensor signal 200 to detect. Furthermore, the control unit 101 set up the signal fault 201 the sensor signal 200 from a measurement error 202 the sensor signal 200 to distinguish. By a reliable distinction between signal interference 201 and measurement errors 202 An increased robustness of the acquisition of the measured data can be provided. In particular, it can be avoided that a short-term signal disturbance 201 leads to an impairment of a vehicle function. Furthermore, a high degree of security can be ensured since actual measurement errors 202 be reliably recognized and taken into account in the provision of vehicle functions.

The control unit 101 is further set up, one or more vehicle functions of the vehicle 100 depending on the sensor signal 200 to control. In particular, the control unit 101 be set up, a control unit 103 to provide a vehicle function (eg, to provide an electronic stability program) to interrupt the vehicle function when it has been detected that the sensor signal 200 a measurement error 202 includes.

As stated above, the control unit is 101 set up a signal fault 201 in the sensor signal 200 to detect. For this purpose, the control unit 101 be set up in that 4 illustrated method 400 perform. The procedure 400 includes determining 401 a reference signal from the sensor signal 200 , 3 shows an exemplary sensor signal 200 and one from the sensor signal 200 determined reference signal 300 , The reference signal 300 can be determined in particular by means of a Kalman filter. This can cause phase distortion between the reference signal 300 and the sensor signal 200 be kept low. In other words, the use of a Kalman filter allows timely detection of signal interference 201 based on the reference signal 300 , Furthermore, the use of a Kalman filter allows one in the sensor signal 200 implicitly account for the noise component (via a covariance matrix of the sensor signal determined with the Kalman filter) 200 ).

A Kalman filter explicitly distinguishes between the dynamics of the system state (eg, a state of the vehicle 100 passing through the sensor 102 to be detected) and the process of measuring the state of the system (ie, detecting the state of the vehicle 100 through the sensor 102 ). The condition of the vehicle 100 At time k, X _k can be described by a multi-dimensional variable (eg, by a vector). The condition of the vehicle 100 with respect to one through the sensor 102 detected measured variable can, for. B. by one-dimensional values x _k , x _k-1 , ..., x _{k-L + 1 of} the measured variable at different sampling times k, k - 1, ..., k - L + 1 are reproduced. The sequence of values _xk , _xk-1 , ..., _{xk-L + 1} describes the sensor signal 200 ,

The present document proposes the state X _{k of} the vehicle 100 by the value of the detected measured variable x _k at the time k, as well as by one or more derivatives of the measured variable at the time k to describe. As stated above, the measured variable x may be a yaw angle of the vehicle 100 act. The condition of the vehicle 100 at time k with respect to a yawing motion can be determined by the yaw angle x _k , by a first derivative of the yaw angle x. _k (ie by a degree of change of the yaw angle or by a yaw rate), by a second derivative of the yaw angle x .. _k (ie by a yaw acceleration) and / or by a third derivative of the yaw angle x ... _k (ie by a Yaw pressure). The state vector X _{k of} the vehicle 100 at time k can thus be described by the value of the measurand, as well as by values for one or more derivatives of the measurand, ie X _k = (x _k , x _k , x .. _k , x ... _k , ... ) ^T. By way of example, it is assumed below that the state vector is the measured variable (ie the sensor signal 200 ), as well as the first three derivatives of the measured variable (ie the first three derivatives of the sensor signal 200 ). As explained above, the derivatives correspond to the velocity of the measured variable, the acceleration of the measured variable and a jerk of the measured variable.

By means of a transition matrix F _k , the transition from a past state to a current state can be determined, ie X _k = F _k-1 X _k-1 + W _k-1 . The transition matrix F _k can be described or modeled by a system model. W _k-1 describes a noise component of the system and can be called system noise. In this document, it is proposed to describe a measured vehicle size by a polynomial. In particular, it is proposed the course of the sensor signal 200 (eg the course of the measured yaw angle of the vehicle 100 ) by a polynomial. It can be considered that the polynomial model is a sign change of a slope of the sensor signal 200 should allow to describe transitions from rising to falling readings (and vice versa). Such sign changes are made possible, in particular, by polynomials of the 3rd and higher order. An exemplary system model is given by a 3rd order polynomial.

The state X _{k of} the system can be detected by the sensor 102 be recorded. This measuring process can be described by an observation matrix H _k . The measured value Z _k at time k then results as Z _k = H _k X _k + V _k , where V _{k represents} the measurement noise. Measurement noise and system noise can typically be fully described by their mean and covariance. In the present document, it is proposed to describe the observation matrix H _k such that the measured value Z _k at the time k corresponds to the value x _{k of} the measured variable, ie

Kalman filtering typically involves two steps. In a first step, the transition matrix F _{k-1 is used to determine} an a priori estimated value of the vehicle's current state, where X _k = F _k-1 X ∨ _k-1 , where X _{k is} the a priori estimate of the current state at time k and X ∨ _{k-1 is} the a posteriori estimate of the previous state at time k-1. It has been shown that due to the chosen polynomial model, the transition matrix F converges to a constant matrix. This transition matrix can be found in Advance analytical calculated and / or determined experimentally.

An estimate error y _k can be determined from the a priori estimated value X _{k of} the current state, as a difference between the actually measured value x _{k of} the measured variable and the a priori estimated value x _{k of} the measured variable, ie y _k = x _k -x ^ _k , where the a priori estimated value x ^ _{k of} the measured variable corresponds to the first entry of the a priori estimated value X ^ _{k of} the current state.

The a posteriori estimate x ∨ _{k of} the measurand at time k can be determined by the following formula: x ∨ _k = x ^ _k + ky _k . Here, K is the so-called Kalman gain, which can be determined from the transition matrix, the observation matrix and the covariance of the state noise and the measurement noise. Like the transition matrix, the Kalman gain can also be determined in advance. This enables efficient calculation of the reference signal 300 from the sensor signal 200 during operation of the vehicle 100 ,

By means of the Kalman filter, it is thus possible to derive from a sequence of values x _{k of} the measured variable (ie from the sensor signal 200 ) a sequence of a posteriori estimates x ∨ _{k of} the measured variable can be determined. This sequence of a posteriori estimates x ∨ _k can be used as a reference signal 300 be used.

In relation to 4 Now, further steps of the exemplary method 400 for the detection of a signal disturbance 201 in the sensor signal 200 described. The procedure 400 includes further, comparing 402 the sensor signal 200 with the reference signal 300 , In particular, a deviation ε _k = x _k -x ∨ _k at the time k between the sensor signal 200 and the reference signal 300 be determined. Furthermore, it can be determined (step 403 ), whether the deviation ε _{k is} equal to or greater than a threshold value. In particular, it can be determined whether the absolute value | ε _k | the deviation is equal to or greater than the threshold. This can be an indication of the presence of a signal disturbance 201 and / or for the occurrence of a measurement error 202 be.

A deviation between sensor signal 200 and reference signal 300 can be due to noise of the sensor signal 200 caused. As is the energy of the noise of the sensor signal 200 Over time, it is beneficial to have the energy of the noise of the sensor signal 200 in the investigation 403 an indication of the presence of a signal disturbance 201 and / or for the occurrence of a measurement error 202 to take into account. This can be a value for the energy of the noise of the sensor signal 200 be determined (eg by isolation of high-frequency components of the sensor signal 200 ). Furthermore, the threshold of the determined. Value for the energy of the noise can be made dependent.

The procedure 400 can continue to disambiguate 404 between the presence of a signal fault 201 and the occurrence of a measurement error 202 include. For this purpose, it can be determined in particular whether, in a predefined period of time after the point in time k at which the indication of the presence of a signal disturbance 201 and / or for the occurrence of a measurement error 202 was determined, a further deviation ε _{k + l is} determined whose absolute value reaches or exceeds the threshold value. In this case, 1 ≦ L, where L corresponds to a predefined number of sampling times, and thus defines the predefined time period after the time k. Furthermore, it can be determined whether the further deviation ε _{k + 1 has} the same or an inverse sign than the first deviation ε _k . If, during the predefined time interval (eg 50 ms), a second deviation ε _{k + 1 is} detected, which reaches or exceeds the threshold value, and which has an inverse sign than the first deviation ε _k , it can be concluded that a signal interference 201 is present. On the other hand, the presence of a measurement error can 202 getting closed.

Thus, within the predefined time interval after detection of a significant deviation ε _{k, it} can be determined whether a short-term signal disturbance 201 or if a measurement error 202 is present. In the first case can be waived, one of the sensor signal 200 restrict or switch off the dependent vehicle function. Thus, an increased continuity of vehicle functions can be provided. In the second case, the vehicle function can be switched off or reduced in scope. It can thus ensure the safety of the vehicle 100 increase.

The present document describes a method and a corresponding control unit which make it possible to detect short-term signal disturbances in a robust, timely and reliable manner. This provides an improved solution to the trade-off between robustness and security. Short term signal disturbances can be detected with increased reliability and disambiguated from the presence of measurement errors. As a result, unnecessary shutdowns of vehicle functions can be avoided. This increases the availability of the corresponding vehicle functions and thus vehicle safety and / or vehicle comfort.

The present invention is not limited to the embodiments shown. In particular, it should be noted that the description and figures are intended to illustrate only the principle of the proposed methods, apparatus and systems.

Claims (10)

Procedure ( 400 ) for detecting a signal disturbance ( 201 ) of a sensor signal ( 200 ), wherein the sensor signal ( 200 ) comprises a sequence of measurements x _k to a corresponding sequence of times k; the method ( 400 ), - determining ( 401 ) of a reference signal ( 300 ) by applying a Kalman filter to the sensor signal ( 200 ); where the reference signal ( 300 ) comprises a sequence of reference values; and - detecting ( 402 . 403 . 404 ) of a signal disturbance ( 201 ) of the sensor signal ( 200 ) based on the reference signal ( 300 ). Procedure ( 400 ) according to claim 1, further comprising, - determining ( 402 ) a deviation ε _k between a measured value x _k and a corresponding reference value x ∨ _k ; - Determine ( 403 ) an indication of a signal disturbance ( 201 ) or for a measuring error ( 202 ) of the sensor signal ( 200 ) at time k, on the basis of the determined deviation ε _k ; and - disambiguating ( 404 ) between a signal disturbance ( 201 ) or a measurement error ( 202 ) of the sensor signal ( 200 ) at time k, based on the reference signal ( 300 ). Procedure ( 400 ) according to claim 2, wherein said determining ( 403 ) of an indication, determining that an amount of the determined deviation ε _{k is} equal to or greater than a threshold value. Procedure ( 400 ) according to claim 3, wherein the threshold value of an energy of a noise component of the sensor signal ( 200 ) depends. Procedure ( 400 ) according to one of claims 2 to 4, wherein the disambiguation ( 404 ), determining a further deviation ε _{k + 1} between a measured value x _{k + 1} and a corresponding reference value x ∨ _{k + 1} at a time k + 1, which lies after the time k at which the indication for a signal disturbance ( 201 ) or for a measuring error ( 202 ) of the sensor signal ( 200 ) was determined; - Determine a further indication of a signal disturbance ( 201 ) or for a measuring error ( 202 ) of the sensor signal ( 200 ) at time k + 1, based on the further deviation ε _{k + 1} ; and - disambiguating ( 404 ) between a signal disturbance ( 201 ) or a measurement error ( 202 ) of the sensor signal ( 200 ) on the basis of the deviation ε _k and on the basis of the further deviation ε _{k + l} Procedure ( 400 ) according to claim 5, wherein the disambiguating ( 404 ), - determining whether the deviation ε _{k has} a different sign than the further deviation ε _{k + 1} ; and / or determining whether a time interval between the time k at which the signal interference indicator ( 201 ) or for a measuring error ( 202 ) of the sensor signal ( 200 ) and the time k + 1 at which the further indication for a signal disturbance ( 201 ) or for a measuring error ( 202 ) of the sensor signal ( 200 ) is equal to or less than a predefined period of time. Procedure ( 400 ) according to any preceding claim, wherein - the sensor signal ( 200 ) from a sensor ( 102 ) of a vehicle ( 100 ) was recorded; The Kalman filter a state of the vehicle ( 100 ) is described by a state vector X; The state vector X at time k the sensor signal ( 200 ) at time k, and one or more derivatives of the sensor signal ( 200 ) at time k; and - a state transition of the vehicle ( 100 ) between states at different times by a third or higher order polynomial. Procedure ( 400 ) according to any preceding claim, wherein the sensor signal ( 200 ) Measured values x _{k of} a yaw sensor ( 102 ). Control unit ( 101 ) for a vehicle ( 100 ), where the vehicle ( 100 ) a sensor ( 102 ) for detecting a sensor signal ( 200 ), wherein the sensor signal ( 200 ) comprises a sequence of measurements x _k to a corresponding sequence of times k; the control unit ( 101 ), - a reference signal ( 300 ) by applying a Kalman filter to the sensor signal ( 200 ) to investigate; where the reference signal ( 300 ) comprises a sequence of reference values x ∨ _k ; and - a signal disturbance ( 201 ) of the sensor signal ( 200 ) based on the reference signal ( 300 ) to detect. Control unit ( 101 ) according to claim 9, wherein the control unit ( 101 ), - based on the reference signal ( 300 ), a signal disturbance ( 201 ) of the sensor signal ( 200 ) of a measuring error ( 202 ) of the sensor signal ( 200 to disambiguate; To cause a functional range of a vehicle function to transmit the sensor signal ( 200 ), is reduced when a measurement error ( 202 ) of the sensor signal ( 200 ) is recognized; and or - to cause the functionality of the vehicle function, which the sensor signal ( 200 ) is maintained, when a signal disturbance ( 201 ) of the sensor signal ( 200 ) is recognized.

Images