Method for measuring the variance in a measurement signal, method for data fusion, computer program, machine-readable storage medium, and device

Abstract

The disclosure relates to a method for measuring the variance in a measurement signal, comprising the following steps: filtering the measurement signal by means of a high-pass filter in order to obtain a filtered measurement signal; determining the variance by using the filtered measurement signal.

Claims

1. A method for measuring a variance of a measurement signal, the method comprising: receiving the measurement signal from a sensor; filtering the measurement signal to obtain a filtered measurement signal by filtering the measurement signal first with a high pass filter and subsequently with a low pass filter; and determining the variance of the measurement signal by filtering a running mean value of the filtered measurement signal.

2. The method as claimed in claim 1, wherein the high pass filter is a linear phase filter.

3. The method as claimed in claim 2, wherein coefficients of the high pass filter are dependent on a sampling rate with which the measurement signal is detected.

4. The method as claimed in claim 2, further comprising: adapting a group delay of the high pass filter to a delay in the variance relevant to a measurement task.

5. The method as claimed in claim 1, further comprising: providing the variance of the measurement signal to a Kalman filter configured to fuse the measurement signal with a further input signal.

6. The method as claimed in claim 1, wherein cutoff frequencies of at least one of the high pass filter and the low pass filter are based on a frequency at which information relevant to a measurement task is still present in the measurement signal.

7. The method as claimed in claim 1, the determining of the variance further comprising: calculating the variance using at least one of a running calculation, a running mean, and a running mean of a sum of squares.

8. The method of claim 1, wherein the method is executed by a non-transitory computer program.

9. The method of claim 8, wherein the non-transitory computer program is stored on a computer-readable storage medium.

10. The method as claimed in claim 2, wherein the high pass filter is a Finite Impulse Response filter.

11. The method as claimed in claim 4, the adapting further comprising: adapting the group delay to be between 50 ms and 100 ms.

12. The method as claimed in claim 11, the adapting further comprising: adapting the group delay to be 80 ms.

13. The method as claimed in claim 6, wherein the cutoff frequencies are located between 2 Hz and 20 Hz.

14. The method as claimed in claim 13, wherein the cutoff frequencies are located at 5 Hz and 10 Hz.

15. A method for data fusion using a Kalman filter, the method comprising: receiving, with the Kalman filter, a measurement signal from a sensor as a first input signal and a variance of the measurement signal as a second input signal, wherein the variance of the measurement signal is measured by (i) filtering the measurement signal to obtain a filtered measurement signal by filtering the measurement signal first with a high pass filter and subsequently with a low pass filter, and (ii) determining the variance of the measurement signal by filtering a running mean value of the filtered measurement signal; and fusing the measurement signal with at least one third input signal using the Kalman filter.

16. The method as claimed in claim 15, wherein the variance of the measurement signal is measured outside of the Kalman filter.

17. A device for measuring a variance of a measurement signal, the device comprising: an input configured to receive the measurement signal from a sensor; and a filter having a high pass filter and a low pass filter, wherein the filter is configured to (i) obtain a filtered measurement signal by filtering the measurement signal first with the high pass filter and subsequently with the low pass filter and (ii) determine the variance by filtering a running mean value of the filtered measurement signal.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) In the following, embodiments of the present disclosure are presented and explained based on the drawings. Shown are:

(2) FIG. 1 a block circuit diagram of an embodiment of the present disclosure;

(3) FIG. 2 a graph with input signals 1 to 4;

(4) FIG. 3a a graph with variance calculations based on input signal 1 with different methods;

(5) FIG. 3b a graph with variance calculations based on input signal 2 with different methods;

(6) FIG. 3c a graph with variance calculations based on input signal 3 with different methods;

(7) FIG. 3d a graph with variance calculations based on input signal 4 with different methods;

(8) FIG. 4 a block circuit diagram;

(9) FIG. 5 a flow diagram of an embodiment of the method for measuring the variance in a measurement signal according to the present disclosure;

(10) FIG. 6 a flow diagram of an embodiment of a method for data fusion according to the present disclosure.

DETAILED DESCRIPTION

(11) FIG. 1 shows a block circuit diagram of an embodiment of the present invention. The block diagram clearly shows the core of the present invention. Sensors S deliver sensor signals or measurement signals to a post-processing unit. For the data fusion of sensor signals it is advantageous to use a Kalman filter K, which is applied to the measurement signals. For this purpose, the measurement signals are fed on the one hand to the Kalman-filter K, and the measurement signals are also fed to a filter, here a high-pass filter HP, to suppress the DC component, hence the real signal. This filtered measurement signal is then fed to the variance calculation or measurement. The measured variance is in turn fed to the Kalman filter K and then evaluated as a further input variable to the data fusion.

(12) FIG. 2 shows a graph with four measurement signals (signal 1 to 4). As can be seen from the graph, the measurement signals differ in their amounts of variance. In the following figures, in other graphs the results of various methods for determining the variance are shown in comparison to an embodiment of the method of the present disclosure.

(13) FIGS. 3a to 3d show graphs with results of the variance calculation in accordance with the window method and a pure low-pass filtering, in comparison to an embodiment of the method 500 of the present disclosure.

(14) The results clearly show the power of the disclosure described, since the results of the embodiment of the method of the present disclosure vary in a much narrower range about the reference variable, which is designated as the input variance.

(15) It is therefore clear that the present disclosure is applicable to a very wide range of input signals and delivers good results.

(16) FIG. 4 shows a block circuit diagram of an embodiment of a system having a device according to the present disclosure.

(17) FIG. 4 shows two sensors S1 and S2, the sensor signals of which, and hence their measurement signal, are fused by means of a Kalman filter K.

(18) To achieve this, the sensors S1, S2 input their measurement signal, on the one hand, directly into the Kalman-filter K as an input signal. In addition, the measurement signals are filtered in accordance with the method 500 of the present invention by means of a high-pass filter HP. The measurement signal filtered in this way is then filtered using a low-pass filter LP. The result of this filter step is input as a (measured) variance of the respective measurement signal into the Kalman filter K as an additional input variable.

(19) The combination of high-pass and low-pass filter is also referred to as a band-pass filter BP. Thus, as an alternative to two individual filters a band-pass filter BP can also be used.

(20) It goes without saying that the filters HP, LP, BP can be designed in different ways. The filters can be implemented in hardware or software, or as a combination thereof.

(21) The resulting output of the Kalman filter K is the fused result of the two measurements or sensor signals.

(22) In the field of driving dynamics of road vehicles the relevant information can be found in the signal between 3 Hz to 20 Hz, in particular from 5 Hz to 20 Hz.

(23) These boundary conditions can be used to derive the result for the high-pass filter HP that the cutoff frequency of the low-pass filter LP, from which the high-pass filter HP can be generated (e.g. by inversion), should lie between approximately 5 Hz and 10 Hz, because due to the minimum possible group delay the damping will not be very high, even up to 20 Hz. The low-pass filter LP should have a cutoff frequency of at least 2 Hz, also in order not to contribute an excessively high value to the group delay.

(24) For use in motor vehicles, a 16th order FIR high-pass filter with the coefficients b.sub.HP=[1, 16, 36, 55, 73, 84, 93, 102, −920, 102, 93, 84, 73, 55, 36, 16, 1] and a.sub.HP=1024 and an infinite impulse response (IIR) low-pass filter with the coefficients b.sub.LP=[1] and a.sub.LP=[16, −15] are proposed for a sampling rate of 200 Hz. This achieves a group delay in the passband of approximately 8 to 15 samples. At a sampling rate of 200 Hz this corresponds to a group delay of 40 ms to 75 ms.

(25) FIG. 5 shows a flow diagram of an embodiment of a method for measuring the variance in a measurement signal according to the present disclosure.

(26) In step 501, the measurement signal is filtered using a high-pass filter in order to obtain a filtered signal.

(27) In step 502, on the basis of the filtered measurement signal the variance in the measurement signal is determined.

(28) FIG. 6 shows a flow diagram of an embodiment of a method for data fusion in accordance with the present disclosure.

(29) In step 601, input signals are fused by means of a Kalman filter, wherein for the determination of the variance of the input signals a method for measuring the variance in a measurement signal according to the present disclosure is applied.