DE102018203802A1 - Method, device and means of transport for multi-stage filtering of a signal in the electrical system of a means of transport - Google Patents

Abstract

The invention relates to a method for the multi-stage filtering of a first signal (S) in the on-board network of a means of locomotion, comprising the steps of: filtering the first signal (S) by means of a first Kalman filter (46) representing a first stage (42) for generating a second signal (S ') and filtering the second signal (S') by means of a second Kalman filter (48) representing a second stage (44) for generating a third signal (S ").

Description

The present invention relates to a method, a device and a means of locomotion for multi-stage filtering of a first signal in the electrical system of a means of locomotion. In particular, the invention relates to a time-efficient filtering associated with an effective smoothing behavior based on a multi-stage Kalman filter.

Discrete-time, one-stage Kalman filters are known in the art, including: be used for position determination in GPS-based navigation systems or systems for environment detection of means of locomotion, for example, to filter noisy sensor signals in the electrical system of the means of transport or to estimate replacement values for faulty or failed sensor signals. By means of a Kalman filter, the states of a noisy, technical process can be optimally estimated. For this, the Kalman filter minimizes the mean square error of the estimate. A prerequisite for the realization of the Kalman filter is a sufficiently accurate, mathematical description of a state space model of a technical process or system in the form of a time-discrete difference equation system. The future states estimated by the Kalman filter are each corrected on the basis of current, real measured values. With an increasing number of real measured values over time, the estimate approaches the actual states theoretically as much as desired.

In addition, due to its mathematical structure, the Kalman filter can be implemented relatively simply on a digital computer and is suitable for use in real-time systems due to a moderated resource requirement in this context. Time-continuous Kalman filters are also feasible, but they are of little practical importance because they are mathematically much more complex and there are few known solutions to this type of Kalman filter.

In connection with a means of transportation, discrete-time Kalman filters are suitable, for example, for determining a vertical acceleration of a wheel of the means of locomotion. In the prior art acceleration sensors are used for this purpose to provide dynamic / active damper systems of the means of transport with the required motion variables for damping control and roll stabilization. For this purpose, an acceleration sensor is arranged in the region of the chassis and in particular in the area of the unsprung mass of the chassis and outputs direct signals representing the vertical acceleration of the wheel. Vertical acceleration is understood to mean such a motion variable or a corresponding signal which describes the acceleration of the wheel essentially in the direction of the road normal or in the vertical direction. The damper control can make appropriate settings for the damper resistance in response to these signals, for example, to facilitate a deflection of the wheel when it has to deflect due to a bump a long way. The same applies to roll stabilization, except that it moves the wheels vertically. In this way, the driving safety and ride comfort can be significantly improved.

Moreover, due to a need for headlamp leveling and sometimes also for controlling active air suspensions in the prior art, level sensors are used which provide information on how much a particular wheel of a vehicle is currently being deployed. The information of the level sensor can be understood as essentially quasi-static state variables. Depending on this size, the headlight range / the angle at which the light beam is aligned and the height / filling of the air suspension are controlled or regulated. The same applies to coilovers.

To save the acceleration sensor, the height signal of the wheel taken directly from the level sensor can be converted into a vertical acceleration signal of the wheel by differentiating twice over time, which can replace the signal generated by the acceleration sensor of the above-mentioned systems. The twofold differentiation and the smoothing of the signal can be performed on the basis of an analog and / or a digital filtering.

The problem in this context is a sometimes high proportion of unknown influences or disturbances which can superimpose the height signal measured by the level sensor. These superimpositions of an actual useful signal, which are also referred to below as process noise, may possibly lead to malfunctions and / or functional failures in the means of transportation without conditioning the noisy signal.

Depending on a signal-to-noise ratio of the level signal, simple filters with a low filter order may suffice, or even complex filters with a high filter order may be required. For example, the ride height signal in different vehicle types can produce a significantly different signal Have noise ratio. The signal-to-noise ratio can be influenced for example by a specific chassis construction and / or by inductive and / or capacitive interference from the surroundings of the level sensor.

Low noise level signals can i.d.R. be processed by means of simple analog and / or digital filters with relatively low order so that their output signal for the subsequent two-fold differentiation for generating the signal for the vertical acceleration of the wheel is suitable. The higher filter order required for highly noisy signals to smooth the signal can be a problem in that it usually involves a higher propagation delay in providing the filtered vertical acceleration signal.

In particular, in connection with a control and / or regulation of the driving dynamics of a means of transport are usually low propagation delays a signal processing chain essential to avoid any functional degradation and / or malfunction in a filtering downstream part of the signal processing chain. For this reason, strongly noisy signals with a conventional filtering can in many cases only be processed to a limited extent since they do not meet the requirement for effective smoothing of the noisy signal, or at the same time for a short signal propagation time of the filter.

It is therefore an object of the present invention to alleviate the above-mentioned disadvantages of the prior art.

The solution of the object identified above is achieved by the features of the independent claims. The dependent claims have preferred developments of the invention to the content.

According to a first aspect, a method for the multi-stage filtering of a first signal in the on-board network of a means of transportation is proposed, and in particular a multi-stage filtering based on a multi-stage Kalman filter. The means of transportation may be, for example, a road vehicle (e.g., motorcycle, car, van, truck) or an aircraft / aircraft.

The basic structure and operation of a conventional single-stage Kalman filter is well described in the art and has been used successfully for many years in a variety of technical applications, therefore omitting a detailed description of the structure and particularly the mathematical structure of the Kalman filter becomes.

In a first step of the method according to the invention, the first signal is filtered by means of a first Kalman filter representing a first stage for generating a second signal. The first signal may be, for example, an output signal of a measuring sensor of a vehicle electrical system of a means of locomotion, such as the level sensor of the level sensor described above. Furthermore, the output signal of the measuring sensor may be superimposed by undesired interference variables, such as a noise signal. These disturbance variables can be generated, for example, by the sensor itself and / or interspersed in it. As causes here u.a. a thermal noise and / or quantization noise of the sensor and / or the sensor electronics and / or inductive and / or capacitive interference from the environment of the sensor in question. In addition, a type, an arrangement and a length of a cable connection from the sensor to a receiving component of the sensor signals can have a not inconsiderable influence on the type and proportion of disturbances in the actual measurement signal.

The Kalman filter of the first filter stage may be a time-discrete Kalman filter known from the prior art, which is realized on the basis of a computer program which is executed by means of an evaluation unit. The evaluation unit may be designed, for example, as a processor, digital signal processor, analog signal processor, microcontroller, electronic control unit, or the like. Preferably, the evaluation unit is connected to a volatile and a non-volatile memory. Furthermore, the evaluation unit may be a component of an existing driver assistance system or of a further control device of a means of transportation, or an independent control device.

The evaluation unit may have a digital and / or analog interface for receiving the first signal. Depending on the form of provision of the first signal (analog or digital), a conversion to a digital signal may initially be necessary for processing the first signal. This can be done by a corresponding analog / digital (A / D) converter connected to the evaluation unit. For the example of the level signal, a bit depth in the range of 8 bits to 12 bits and in particular of 10 bits has proven to be sufficient when the signal is sampled by an A / D converter. A suitable sampling frequency of the level signal can be, for example, between 100 Hz and 1 kHz and in particular at 400 Hz.
Alternatively, the ride height sensor can already output the ride height signal in digital form and be received by the evaluation unit via a corresponding digital interface. For a through the level sensor digital output level signal also apply the above preferred values for a bit depth and a sampling frequency.

The first signal received by the evaluation unit may represent a single input, such as a measurement such as the altitude signal, or a vector of inputs. In this context, a first element provided in the first stage may be advantageous for a selection and weighting of the input variables. The first element for selection and weighting upstream of the first Kalman filter can be realized as a function of an analog or digital first signal, for example by means of an analog circuit or by means of a program code executed by means of the evaluation unit.

Through the selection, a static selection of predefined input variables adapted to the respective task of the multi-stage Kalman filter can be performed. It is also conceivable to select dynamically between a first predefined combination of input variables and a second predefined combination of input variables that deviates from the first combination. This may be advantageous, for example, when the first signal comprises measured values of a plurality of redundant sensors in order to ensure a higher level of reliability of an overall system. In addition, selecting a specific combination of inputs may also be performed in dependence on predefined environmental conditions and / or events.
A weighting of the individual input variables can likewise be carried out statically or dynamically analogously to the selection of the input variables in order to be able to adapt the input variables, for example, to a specific configuration or task of the filter and / or to react in response to predefined environmental conditions and / or events. In the event that the selection and weighting stage is not required for specific applications of the multi-stage Kalman filter, it may optionally be dispensed with or deactivated by means of a weighting factor of one and a standard forwarding of all input signals.

After the (optional) processing of the first signal by means of the first element for selection and weighting, the first signal can be fed to the filtering by the first Kalman filter. This can be done by processing individual program steps of the computer program within the evaluation unit. The first Kalman filter generates from the first signal a second signal representing an output signal of the first Kalman filter.

In the case of the ride height signal as a first signal to be filtered, a smoothed vertical speed signal and a smoothed vertical acceleration signal of the wheel can be calculated from the ride height signal by suitable dimensioning of the first Kalman filter. The vertical acceleration signal computed by the first stage has a similar delay due to the propagation delay through the first Kalman filter, such as filtering based on conventional filtering techniques, e.g. using a dual differentiator in combination with a FIR filter. The respective propagation delays of the first stage of the multi-stage Kalman filter and the conventional filtering method with a respectively sufficient smoothing behavior are in the order of about 10 ms to 20 ms delay time compared to a directly measured vertical acceleration signal by a vertical acceleration sensor. In many cases, these values may be too high in the context of vehicle dynamics control systems to allow sufficiently fast intervention in vehicle dynamics. This can, for example, have a negative effect on driving safety and / or driving comfort.

In contrast, the vertical speed signal filtered by the first stage of the multi-stage Kalman filter has a much shorter propagation delay compared to the first stage vertical acceleration signal, in which case the vertical speed signal generated in the first stage is preferably provided as a second signal at the output of the first stage.

The method according to the invention provides, in a second step, for filtering the second signal by means of a second Kalman filter representing a second stage in order to generate a third signal. The second stage filter structure of the multi-stage Kalman filter may be identical or different to the first stage filter structure. This also applies to a second element for selection and weighting upstream of the second Kalman filter, which may likewise be identical or different from the first element for selection and weighting. An identical or different design of the filter structures and / or an identical or different design of the element for selection and weighting can be carried out as a function of a corresponding implementation variant (eg analog, digital, circuit technology, programmatic) and in particular as a function of a multi-stage Kalman filter Task defined.

In principle, any desired mixed forms with regard to the realization of the first and second stages are also conceivable. So the first one For example, the stage may be a Kalman filter based on an analog circuit technique and the second stage may be a Kalman filter based on a computer program of a microcontroller, provided A / D conversion is performed between the first and second stages.

The second signal, which represents the vertical speed signal of the first stage, can be converted by the second Kalman filter again with a low propagation delay into a smoothed vertical acceleration signal, which represents the third signal at the output of the second stage. The division of the filtering of the level signal to several filter stages leads to the inventive advantage of filtering with a significantly lower propagation delay compared to a filter with a conventional single-stage Kalman filter or a combination of a conventional dual differentiator with an FIR filter at each similar smoothing behavior.

In the case of the filtering of the level signal by the multi-stage Kalman filter, a propagation delay of a few milliseconds can be achieved, so that, in the context of regulation and / or control of driving dynamics of the means of transport, maximum allowable propagation delays of about 4 ms to 5 ms reach, or even fall below.

The third signal generated by means of the second Kalman filter from the second signal can in turn consist of a single variable representing the third signal or of a vector of individual variables. The third signal may then be provided at the second stage output for subsequent processing.

It should be noted in particular that the inventive multi-stage Kalman filter is not limited to a cascading of a first and a second Kalman filter. Rather, any number of Kalman filters can be cascaded in order to achieve, depending on a specific application, an optimized signal smoothing for this application with the lowest possible propagation delay.

In a further advantageous embodiment of the present invention, in the filtering of the second signal in addition, the first signal in the second Kalman filter incorporated by this processed in the first stage on the one hand by the first Kalman filter and on the other in unchanged form to the Output of the first stage forwarded or looped through. This means that the subsequent second stage can thus access both the first signal and the second signal, which may be advantageous for filtering in the second stage, depending on the application. Since the first signal can consist of a vector of input variables, by an appropriate configuration of the second element for selection and weighting either the entire vector of input quantities of the first signal, or a suitable sub-vector of input quantities of the first signal can be fed into the second Kalman filter become.

As an alternative to looping through the first signal through the first stage, the first signal based on an analog and / or digital circuit implementation can also be transmitted to the second stage by means of parallel signal routing outside the first stage of the multi-stage Kalman filter.

In a multi-stage Kalman filter, which has for example three, four or more filter stages, this principle can be used for each subsequent stage. That specifically, that in a third stage in addition to the third signal (as an output signal of the second stage) in addition, the second and / or the first signal are fed and there a selection and / or weighting can be supplied.

The third signal generated by the second stage, which can represent a derived from the ride height signal, smoothed vertical acceleration signal of the wheel can then be transmitted via the electrical system to a system for controlling and / or regulating a driving dynamics of the means of locomotion and further processed accordingly. In addition, it is also conceivable that in addition to the third signal in addition, the second and / or the first signal to the system for controlling and / or regulating a vehicle dynamics can be transmitted.

In addition, a return of an output signal of a respective stage to the input of one or more preceding stages is conceivable by the respective output signal, or a specific combination of the respective output signal representative individual sizes, in a respective element for the selection and weighting of one or more preceding stages is fed.

Furthermore, between two stages of the multi-stage Kalman filter additional processing of the signal may take place, which is fed from one stage to another stage (in the form of a forward connection between two stages and / or in the form of a recirculated connection). In concrete terms, this means that, for example, the second signal generated by the first stage before the processing by the second stage be subjected to appropriate processing. Such processing can be done, for example, by an additional filter such as a noise reduction filter and / or other filters and / or other signal processing units. In addition, this additional processing can also incorporate additional input variables which do not represent input variables of the respective stages of the multistage Kalman filter. In this context, it should be noted that the propagation delay of the additional signal processing is preferably kept low in order to obtain in particular the inventive advantage of a short processing time by the multi-stage Kalman filter.

In a further advantageous refinement, a first parameter set of the first Kalman filter and / or a second parameter set of the second Kalman filter are adapted in response to a predefined event and / or a driving situation and / or an environmental situation. A preferred realization of this embodiment can again be based on an evaluation unit, which executes a corresponding computer program for implementing the functionality.

In the context of the present invention, the first and second parameter sets are to be understood as meaning a respective complete set of filter parameters for the first and second Kalman filters. It fully describes the filter characteristics of the first and second Kalman filters.

The first and the second parameter set can be represented, for example, by a first and second data record which are stored persistently in the non-volatile memory of the evaluation unit. The computer program for calculating multistage Kalman filtering executed on the evaluation unit can load the data records stored in the non-volatile memory from the non-volatile memory, for example in an initialization phase of the evaluation unit, to obtain the filter structures of the first and second Kalman filters for the Kalman filter initialize subsequent calculation. For this purpose, the parameters of the first and second parameter set can preferably be loaded into a first and second data structure of the volatile memory connected to the evaluation unit for fast data access. A multi-stage Kalman filter initialized in this way is now able to perform the function of the filter determined by the first and second parameter sets.

In response to changing boundary conditions (e.g., environmental influences on a first and / or second signal to be filtered), it may be advantageous if the function to be performed of the multi-stage Kalman filter can be adapted in response to these altered boundary conditions. This can be accomplished, for example, by storing, in addition to the first and second parameter sets, an alternative first and an alternative second parameter set in the non-volatile memory, which differ in each case from the first and second parameter set. Using the example of the regulation and / or control of driving dynamics of the means of locomotion on the basis of the ride height signal, an altered boundary condition may be, for example, a change in a road surface run over by the means of transportation. While the first and second parameter set for Kalman multi-level filtering of the ride height signal may be adapted to the surface of an average road surface, a different surface of a road surface (for example due to a very rough, wavy or damaged surface) may result in significantly poorer filtering results due to the multi-stage Lead Kalman filtering. This can have a correspondingly negative effect on the control and / or regulation of the driving dynamics.

To counteract this problem, the first alternative and the second alternative parameter set may include corresponding configurations that allow adaptation of the multi-level Kalman filtering to, for example, a modified pavement. As soon as such a change of boundary conditions is detected, which can have a negative effect on the filter behavior, the evaluation unit may load the alternative first and / or second parameter set from the non-volatile memory into the location of the first and second data structure in the volatile memory, whereby the first and second parameter set are replaced. The subsequent calculations by the multistage Kalman filter are based on the alternative parameter sets and thus allow for filtering adapted to the respective boundary conditions.

In this way, further alternative parameter sets for a plurality of conceivable changed boundary conditions can be created, which can be adapted accordingly in response to a predefined event and / or a driving situation and / or an environmental situation. The adaptation of the parameter sets can be carried out, for example, in response to an environmental situation and / or driving situation detected by a further sensor and / or by an evaluation of the quality of the output signal of the multi-stage Kalman filter itself. A changed environmental situation can, for example, affect the detection range of an environmental sensor of the means of locomotion, if this is restricted and / or disturbed, for example, by occurring rain or fog.

In addition, a predefined event, such as a user input made via a control element, can also lead to an adaptation of the parameter sets. Thus, for example, a driving mode change (eg change from a comfort mode to a sports mode) of the means of locomotion triggered in this way by the user can bring about an adaptation of the parameter sets in order to realize or support the selected driving mode.

The parameters for weighting and selection describing the first and second elements for selection and weighting can likewise be part of the first and / or second parameter set or of the alternative first and second parameter set. By suitable access to the data structures which represent the parameter sets in the memory, the parameters for weighting and / or selection and / or adaptation of the configuration of the Kalman filters can be adapted individually or in any desired combinations.

The parameter sets can be adapted during operation of the evaluation unit and the components connected to it via the electrical system in the means of transportation. A change in the first and second parameter set of the first and second Kalman filter is usually associated with a transient phase of the multi-stage Kalman filter, whereby the output signal of the filter for a certain period of time after adapting the parameter sets, inaccurate or unusable values. The duration of the transient phase depends on the configuration of the multistage Kalman filter. In connection with the calculation of the vertical acceleration signal from the height signal of the means of locomotion, this lies approximately in a range of 1 ms to 3 ms. Preferably, a receiver of the output signal of the multistage Kalman filter is set up in such a way that it takes into account such an occurring transient phase of the multi-stage Kalman filter with regard to an uninterrupted maintenance of a specific functionality.

In a further advantageous embodiment, the first signal is filtered in parallel by a plurality of different, multi-stage Kalman filters, and further comprising selecting the third signal from a plurality of parallel generated third signals on the basis of predefined criteria. In other words, the first signal may be fed into a plurality of parallel-connected Kalman multistage filters that perform different filtering of the first signal based on the first signal and their respective configuration of the filter parameters. In the context of a processor-assisted realization of Kalman filtering, "arranged in parallel" means that the filterings carried out by means of a computer program can be calculated quasi-parallel one after the other based on the respective first signal. Alternatively or additionally, the calculation for the plurality of parallel-arranged, multi-stage Kalman filters can also be divided into several cores of a multiprocessor system in the sense of a true parallel calculation.

By means of one of the plurality of multi-stage Kalman filter downstream element for a signal selection can be switched between the respective third signals of the plurality of multi-stage Kalman filter. Thus, for example, the different parameter sets adapted to a respective driving mode of the means of locomotion can each be distributed to the plurality of parallel-arranged, multi-stage Kalman filters in order to generate the respective third signals simultaneously. Depending on a user input for a desired driving mode, the downstream element for selection can select the appropriate for the desired driving mode third signal from the plurality of third signals and transmitted to a multi-stage Kalman filter downstream component in the electrical system of the means of transport. It should be noted that the parameter sets of the Kalman filters arranged in parallel do not have to be permanently permanently assigned to the filters. Instead, it is also possible here to exchange the parameter sets of one or more parallel Kalman filters with additional alternative parameter sets.

Another advantage of the Kalman filters arranged in parallel is that the downstream element for selecting the third signal may have enhanced functionality that allows switching from one of the plurality of third signals to another of the plurality of third ones Signals to perform a suitable billing of the two third signals to a resulting third signal. In this way, within a predefined switching period, a smooth transition from one of the plurality of third signals to another of the plurality of third signals can be realized. This may be advantageous for a downstream receiver of the resulting third signal if abrupt switching between two third signals in the receiver results in functional limitations.

In addition, it is also conceivable that the downstream element for selecting the third signal creates a resulting, third signal in the form of a multiplex signal of a plurality of third signals and the receiver of the third signal in this way a plurality of different filter results based on the first signal provides.
A realization of one or more of the above-described configurations of the multi-stage Kalman filter on the basis of a processor, a digital signal processor, or a microcontroller can always be implemented in such a way that it meets the requirements of a real-time system, since the Filter necessary calculation steps are deterministic.

Although the above description of the method according to the invention and the advantageous embodiments for the sake of clarity relate primarily to the application of a filtering a ride height signal of a wheel, the method is not limited to this application. Rather, the multi-stage Kalman filter is capable of improving a variety of prior art applications by replacing a conventional Kalman filter employed therein with the multi-stage Kalman filter of the present invention. The achievable as a function of a specific task, significantly shorter propagation delays of the multi-stage Kalman filter compared to conventional filter methods are particularly in connection with time-critical applications of particular advantage. Thus, for example, the reaction time of existing systems for filtering image signals and / or radar signals and / or ultrasound signals and / or LIDAR signals and / or GPS tracking data in a means of locomotion can be significantly improved by using the multistage Kalman filter. In particular, in connection with driver assistance systems for automated driving and / or for collision avoidance, the use of the multi-stage Kalman filter can thus bring about an improvement in the driving safety or ride comfort desired by these systems.

Moreover, as the example of the vertical acceleration sensor shows, cost savings are possible through the use of the multi-stage Kalman filter, as its filtering properties sometimes allow for elimination or simplification of existing components.
Further, replacement of a conventional Kalman filter by the multi-stage Kalman filter of the present invention is easily possible in many cases, particularly when the conventional Kalman filter is implemented by means of a computer program of a controller executed on a processor. In such a case, the conventional Kalman filter can be replaced by a software update, which is usually inexpensive compared to a hardware exchange, in order to improve the performance of an existing control unit or system.

According to a second aspect, an inventive device for filtering a first signal in the electrical system of a means of transport is proposed. The device comprises an evaluation unit, which may be designed, for example, as a processor, digital signal processor, analog signal processor, microcontroller, electronic control unit, or the like, and may be part of a control device of the means of locomotion. The evaluation unit has a data input and a data output, which are preferably configured in the form of a digital interface. Furthermore, the evaluation unit is set up to filter the first signal in conjunction with the data input by means of a Kalman filter representing a first stage to generate a second signal and the second signal by means of a second Kalman filter representing a second stage to generate a third one To filter signals. The first and the second Kalman filters can be realized, for example, in the form of a computer program executed on the evaluation unit, which performs the individual calculation steps required for the filtering. In addition, a non-volatile memory u.a. be attached to store the filter parameter sets of the multi-stage Kalman filter. Furthermore, a volatile memory u.a. be connected for quick access to the filter parameters that can be transmitted from the non-volatile to the volatile memory, for example, in an initialization phase of the evaluation. Furthermore, the evaluation unit is set up to output the third signal via the data output.

The remarks on the above features can be analogously transferred to the device according to the invention, which is why reference is made to the above statements regarding the advantages, features and combinations of features of the device.

According to a third aspect of the present invention, there is proposed a means of transportation comprising a device according to the second aspect of the invention. The features, combinations of features as well as the advantages resulting therefrom correspond to those embodied in connection with the first and second aspects of the invention so apparent that reference is made to the above statements to avoid repetition.

• 1 ein Flussdiagramm veranschaulichend Schritte eines Ausführungsbeispiels eines erfindungsgemäßen Verfahrens;
• 2 eine schematische Übersicht über Komponenten einer erfindungsgemäßen Vorrichtung in Verbindung mit einem Fortbewegungsmittel;
• 3 ein Blockschaltbild einer erfindungsgemäßen ersten Stufe eines mehrstufigen Kalman-Filters;
• 4 ein Blockschaltbild eines erfindungsgemäßen, mehrstufigen Kalman-Filters;
• 5 ein Beispiel eines mehrstufigen Kalman-Filters zur Berechnung einer Vertikalbeschleunigung eines Rades aus einem Höhenstandssignal des Rades;
• 6 ein Zeitdiagramm zum Vergleich einer zweistufigen Kalman-Filterung mit einer herkömmlichen Filterung mit einem zweifachen Differenzierer in Kombination mit einem FIR-Filter 14-ter Ordnung; und
• 7 ein Zeitdiagramm zum Vergleich einer zweistufigen Kalman-Filterung mit einer herkömmlichen, einstufigen Kalman-Filterung.

Further details, features and advantages of the invention will become apparent from the following description and the figures. Show it:

• 1 a flowchart illustrating steps of an embodiment of a method according to the invention;
• 2 a schematic overview of components of a device according to the invention in conjunction with a means of locomotion;
• 3 a block diagram of a first stage of a multi-stage Kalman filter according to the invention;
• 4 a block diagram of a multi-stage Kalman filter according to the invention;
• 5 an example of a multi-stage Kalman filter for calculating a vertical acceleration of a wheel from a ride height signal of the wheel;
• 6 a timing diagram for comparing a two-stage Kalman filtering with a conventional filtering with a dual differentiator in combination with a 14th order FIR filter; and
• 7 a timing diagram for comparing a two-stage Kalman filtering with a conventional, single-stage Kalman filtering.

1 shows steps of an embodiment of a method according to the invention for the multi-stage filtering of a first signal S in the electrical system of a means of transport 80 , It will be in the first step 100 a first signal S by means of a level sensor 30 a wheel of means of transport 80 generated and by means of a first stage 42 representing the first Kalman filter 46 for generating a second signal S ' filtered. The first Kalman filter 46 has a first set of parameters that describes the features and functionality of the first Kalman filter 46 completely describes. The first parameter set is dimensioned such that by the filtering of the level signal by the first Kalman filter 46 a smoothed signal of a vertical velocity of the wheel is generated, which is the second signal S ' represents. An inventive property of the second signal generated in this way S ' is a significantly reduced propagation delay of the second signal compared to conventional filter methods S ' , At the second step 200 becomes the second signal S ' by means of a second stage 44 representing the second Kalman filter 48 for generating a third signal S " filtered. The second Kalman filter 48 has a second set of parameters that describes the characteristics and operation of the second Kalman filter 48 completely describes. The second parameter set is dimensioned such that by the filtering of the vertical velocity signal by the second Kalman filter 48 , a smoothed signal of vertical acceleration of the wheel is generated, which is the third signal S " represents. An inventive property of the third signal generated in this way S " is again significantly reduced in comparison to conventional filter methods propagation delay of the third signal S " ,

2 shows a schematic overview of components of a device according to the invention in conjunction with a means of transport 80 , The device comprises an evaluation unit 10 which in turn comprises a processor for executing a computer program. The computer program executed on the processor implements a first stage 42 and a second stage 44 a multi-stage Kalman filter for filtering a level signal of a level sensor 30 a wheel of means of transport 80 , Furthermore, the evaluation unit has 10 via a data input 12 , via which they read the level signal of the level sensor 30 via the electrical system of the means of transport 80 in the form of a digital signal. In addition, to the evaluation unit 10 a volatile memory 20 and a non-volatile memory 22 tethered. The non-volatile memory includes, among others, the first Kalman filter 46 descriptive first parameter set and a second Kalman filter 48 descriptive second parameter set. These are in an initialization phase of the evaluation unit 10 in the volatile memory 22 and from there by the computer program for calculating the filtering of the first Kalman filter 46 and the second Kalman filter 48 used. Furthermore, the evaluation unit comprises 10 a data output 14 via which it transmits the altitude signal filtered by the multistage Kalman filter in the form of a vertical acceleration signal of the wheel to a control system for driving dynamics 35 of the means of transport 80 outputs.

und einem Steuerungsvektor $\overrightarrow{u}$

zusammen, welche von einem technischen Prozess 70 stammen. Der Index k kennzeichnet die Werte der Variablen zum Zeitpunkt t_k. Die weiteren Eingangsgrößen ${\overrightarrow{z}}_{\cdot 1}\dots {\overrightarrow{z}}_{\cdot \text{n}-2},{\overrightarrow{z}}_{\cdot \text{n}-1}$

sind Ausgangsgrößen vorangegangener n-1 Kalman-Stufen. Mittels des Elements zur Selektion und Gewichtung 74 wird aus den verfügbaren Eingangsgrößen eine Auswahl tatsächlich benötigter Eingangsgrößen für die Kalman-Filterung getroffen. Die ausgewählten Eingangsgrößen werden anschließend durch das Element zur Selektion und Gewichtung 74 gewichtet und zum Vektor ${\overrightarrow{z}}_{\cdot \text{n*}}$

kombiniert, welcher anschließend dem n-ten Kalman-Filter 49 zugeführt wird. Der Messvektor ${\overrightarrow{z}}_{\cdot 0},$

der Steuerungsvektor $\overrightarrow{u}$

und die Vektoren ${\overrightarrow{z}}_{\cdot 1}\dots {\overrightarrow{z}}_{\cdot \text{n}-2},{\overrightarrow{z}}_{\cdot \text{n}-1}$

werden in der n-ten Kalman-Stufe 45 sowohl durch das n-te Kalman-Filter 49 verarbeitet, als auch an den Ausgang der n-ten Stufe 45 unverändert durchgeschleift und bilden in Kombination mit dem in der n-ten Stufe 45 erzeugten Vektor ${\overrightarrow{z}}_{\cdot \text{n}}$

das Ausgangssignal S_A der n-ten Stufe 49. 3 shows a block diagram of an n-th stage according to the invention 45 a multi-stage Kalman filter in which the nth stage 45 an input signal to be filtered S _E receives. The structure of the illustrated nth stage 45 is exemplary of any stage of the multi-stage Kalman filter, since the basic structure of the different stages is preferably identical. The nth stage according to the invention 45 consists of an nth Kalman filter 49 , an element for selection and weighting 74 and an output matrix Ỹ _n together. The input signal S _E is made up of a measuring vector ${\overrightarrow{z}}_{\cdot 0}$

and a control vector $\overrightarrow{u}$

together, which of a technical process 70 come. The index k identifies the values of the variables at time t _k . The other input variables ${\overrightarrow{z}}_{\cdot 1}...{\overrightarrow{z}}_{\cdot \text{n}-2}.{\overrightarrow{z}}_{\cdot \text{n}-1}$

are output quantities of previous n-1 Kalman stages. By means of the element for selection and weighting 74 will be out of the available inputs a selection of actually required inputs for the Kalman filtering made. The selected input variables are then passed through the element for selection and weighting 74 weighted and to the vector ${\overrightarrow{z}}_{\cdot \text{n *}}$

which then follows the nth Kalman filter 49 is supplied. The measurement vector ${\overrightarrow{z}}_{\cdot 0}.$

the control vector $\overrightarrow{u}$

and the vectors ${\overrightarrow{z}}_{\cdot 1}...{\overrightarrow{z}}_{\cdot \text{n}-2}.{\overrightarrow{z}}_{\cdot \text{n}-1}$

be in the nth Kalman level 45 both through the nth Kalman filter 49 processed as well as to the output of the nth stage 45 passed through unchanged and form in combination with that in the nth stage 45 generated vector ${\overrightarrow{z}}_{\cdot \text{n}}$

the output signal S _A the nth level 49 ,

und einem Steuerungsvektor $\overrightarrow{u}$

zusammen, welches nach einer Selektion und Gewichtung des Messvektors durch das erste Kalman-Filter 46 gefiltert wird. Das Signal S resultiert nach der Filterung in einem zweiten Signal S', welches sowohl die Eingangsgrößen der ersten Stufe 42, als auch den durch das erste Kalman-Filter 46 erzeugten Vektor ${\overrightarrow{z}}_{\cdot 1}$

enthält. Das zweite Signal S' wird anschließend einer zweiten Stufe 44 des mehrstufigen Kalman-Filters zugeführt und dort analog zur ersten Stufe durch das zweite Kalman-Filter 48 gefiltert. Das zweite Signal S' resultiert nach der Filterung durch das zweite Kalman-Filter 48 in einem dritten Signal S", welches sowohl die Eingangsgrößen der zweiten Stufe, als auch den durch das zweite Kalman-Filter 48 erzeugten Vektor ${\overrightarrow{z}}_{\cdot 2}$

enthält. Analog zur Verarbeitung in der ersten Stufe 42 und der zweiten Stufe 44 des erfindungsgemäßen, mehrstufigen Kalman-Filters, kann das dritte Signal S" zur Filterung in eine weiter, n-te Stufe 45 mit einem n-ten Kalman-Filter 49 und ggf. weiteren kaskadierten Stufen zugeführt werden. 4 shows a block diagram of a multi-stage Kalman filter according to the invention, which consists of several cascaded stages of the in 3 described n-th stage 45 composed. A first step 42 The multi-stage Kalman filter includes a first Kalman filter 46 and an item (not shown) for selection and weighting 49 and receives a first signal S , The first signal S is made up of a measuring vector ${\overrightarrow{z}}_{\cdot 0}$

and a control vector $\overrightarrow{u}$

which after selection and weighting of the measurement vector by the first Kalman filter 46 is filtered. The signal S results after filtering in a second signal S ' , which contains both the input variables of the first stage 42 , as well as the first Kalman filter 46 generated vector ${\overrightarrow{z}}_{\cdot 1}$

contains. The second signal S ' then becomes a second stage 44 fed to the multi-stage Kalman filter and there analogous to the first stage by the second Kalman filter 48 filtered. The second signal S ' results after filtering by the second Kalman filter 48 in a third signal S " which includes both the second stage input and the second Kalman filter 48 generated vector ${\overrightarrow{z}}_{\cdot 2}$

contains. Analogous to the processing in the first stage 42 and the second stage 44 the multi-stage Kalman filter according to the invention, the third signal S " for filtering in a further, nth stage 45 with an nth Kalman filter 49 and possibly further cascaded stages are supplied.

einfließen. Aus diesem Grund umfasst ein in die erste Stufe 42 des mehrstufigen Kalman-Filters einfließendes erstes Signal S hier ausschließlich den durch den Höhenstandssensor 30 ermittelten, verrauschten Messwert für einen durch das Rad in vertikaler Richtung zurückgelegten Weg s_HSS . Nach der Filterung des ersten Signals S durch die erste Stufe 42 stehen am Ausgang der ersten Stufe 42 die gefilterten Größen für den zurückgelegten Weg des Rades s₁ , die Vertikalgeschwindigkeit v₁ und die Vertikalbeschleunigung a₁ des Rades in Form eines zweiten Signals S' zur Verfügung. Durch die oben beschriebenen vorteilhaften Eigenschaften des mehrstufigen Kalman-Filters steht des Vertikalgeschwindigkeitssignal v₁ am Ausgang der ersten Stufe 42 mit einer sehr geringen Laufzeitverzögerung zur Verfügung. Aus diesem Grund wird ausschließlich das Vertikalgeschwindigkeitssignal v₁ des Signals S' in die zweite Stufe 44 eingespeist. Dies wird mittels eines Elements zur Selektion und Gewichtung 74 in der zweiten Stufe 44 erreicht, das nur das Vertikalgeschwindigkeitssignal v₁ zum zweiten Kalman-Filter 48 der zweiten Stufe 44 passieren lässt. In der zweiten Stufe 44 erfolgt nun die Filterung des Vertikalgeschwindigkeitssignals v₁ . Am Ausgang der zweiten Stufe 44 stehen anschließend die gefilterten Größen für den zurückgelegten Weg des Rades s₂ , die Vertikalgeschwindigkeit v₂ und die Vertikalbeschleunigung a₂ in Form eines dritten Signals S" zur Verfügung. Das gewünschte Signal a₂ , welches mit hinreichender Genauigkeit der tatsächlichen bzw. einer direkt gemessenen Vertikalbeschleunigung des Rades a_RBS entspricht, wird anschließend an ein Steuerungs- und/oder Regelungssystem für eine Fahrdynamik 35 des Fortbewegungsmittels 80 zur weiteren Verarbeitung übertragen. 5 shows an example of a multi-stage Kalman filter for calculating a vertical acceleration of a wheel from a ride height signal of the wheel. One of a noise 72 superimposed technical process 70 Here is a vertical movement of the wheel of a means of locomotion. An unknown, vertical force acts on the wheel F _wheel , This force changes the vertical motion of the wheel, such as the distance traveled, its speed and acceleration. Because the force F _wheel is unknown in this example, and thus is not present as a stand-alone measurement signal, it also can not function as a control vector supporting Kalman filtering $\overrightarrow{u}$

incorporated. For this reason, one includes in the first stage 42 the multi-stage Kalman filter flowing first signal S here exclusively by the level sensor 30 determined, noisy measured value for a path traveled by the wheel in the vertical direction s _HSS , After filtering the first signal S through the first stage 42 stand at the exit of the first stage 42 the filtered sizes for the distance traveled by the wheel s ₁ , the vertical speed v ₁ and the vertical acceleration a ₁ of the wheel in the form of a second signal S ' to disposal. Due to the above-described advantageous properties of the multi-stage Kalman filter is the vertical speed signal v ₁ at the exit of the first stage 42 with a very low propagation delay available. For this reason, only the vertical speed signal is used v ₁ the signal S ' in the second stage 44 fed. This is done by means of an element for selection and weighting 74 in the second stage 44 achieved that only the vertical speed signal v ₁ to the second Kalman filter 48 the second stage 44 lets happen. In the second stage 44 Now the filtering of the vertical speed signal takes place v ₁ , At the output of the second stage 44 are then the filtered sizes for the distance traveled by the wheel s ₂ , the vertical speed v ₂ and the vertical acceleration a ₂ in the form of a third signal S " to disposal. The desired signal a ₂ which with sufficient accuracy of the actual or a directly measured vertical acceleration of the wheel a _RBS is then connected to a control and / or regulation system for a vehicle dynamics 35 of the means of transport 80 for further processing.

6 FIG. 12 is a timing diagram for comparing a two-stage Kalman filtering with a conventional two-way differentiator filtering in combination with a 14th order FIR filter. FIG. The curve shown by a solid line corresponds to a vertical acceleration signal based on the conventional filtering from a level signal of a wheel of a vehicle 80 was determined. The dotted line curve corresponds to a vertical acceleration signal obtained from the same level signal based on the multi-level Kalman filter. There is a clear difference between the two signals with regard to their propagation delays. The conventional filtering has a significant propagation delay in a smoothing behavior comparable to the two-stage Kalman filter compared to the two-stage Kalman filter.

7 shows a timing diagram for comparing a two-stage Kalman filtering with a conventional, single-stage Kalman filtering. The curve represented by a solid line corresponds to a vertical acceleration signal of a wheel of a traveling means measured by an acceleration sensor. The curve represented by a dotted line corresponds to a vertical acceleration signal which was determined using a two-stage Kalman filter according to the invention from a measured ride height signal of the wheel. The curve represented by a dashed line corresponds to a vertical acceleration signal which was determined using a conventional Kalman filter from the measured ride height signal of the wheel. It can also be seen from the diagram that the measured vertical acceleration signal and the vertical acceleration signal determined by the two-stage Kalman filter have only minimal temporal deviations in their curve progress compared to the vertical acceleration signal determined by the conventional Kalman filter. Due to a Kalman filter underlying signal prediction, it is also possible that the waveform of the two-stage Kalman filter at some times precedes the measured signal.

LIST OF REFERENCE NUMBERS

10

evaluation

12

data input

14

data output

20

volatile memory

22

non-volatile memory

30

Level sensor

35

Control and / or regulation system for driving dynamics

40

Multi-stage Kalman filter

42

first stage

44

second step

45

nth stage

46

first Kalman filter

48

second Kalman filter

49

nth Kalman filter

70

technical process

72

sough

74

Element for selection and weighting

80

Means

S

first signal

S '

second signal

S "

third signal

S _E

input

S _A

output

s ₁ , s ₂

path

v ₁ , v ₂

speed

a ₁ , a ₂

acceleration

100, 200

steps

Claims (12)

Method for multi-stage filtering of a first signal (S) in the electrical system of a means of transport (80) comprising the steps: Filtering the first signal (S) by means of a first Kalman filter (46) representing a first stage (42) for generating a second signal (S ') and Filtering the second signal (S ') by means of a second Kalman filter (48) representing a second stage (44) for generating a third signal (S "). Method according to Claim 1 wherein the filter structures of the first Kalman filter (46) and the second Kalman filter (48) are identical or different. Method according to one of the preceding claims, wherein the individual stages (42, 44, 45) of the multi-stage Kalman filter (46, 48, 49) are implemented in analog and / or digital form. Method according to one of the preceding claims, wherein in the filtering of the second signal (S ') additionally the first signal (S) flows into the second Kalman filter (48). Method according to one of the preceding claims, wherein The first signal (S) is an altitude signal of a wheel of the means of locomotion (80), • the second signal (S ') is a signal of a vertical speed of the wheel, and • the third signal (S ") is a signal of a vertical acceleration of the wheel. Method according to one of the preceding claims, wherein the first signal (S) and / or the second signal (S ') and / or the third signal (S ") in a system for controlling and / or regulating a driving dynamics (35) of the means of locomotion (80) flows in. Method according to one of the preceding claims further comprising adapting a first parameter set of the first Kalman filter (46) and / or a second parameter set of the second Kalman filter (48) in response to a predefined event and / or a driving situation and / or an environmental situation. The method of any one of the preceding claims, wherein the first signal (S) is filtered in parallel by a plurality of different multi-stage Kalman filters, and further comprising selecting the third signal (S ") from a plurality of parallel generated third signals according to predefined criteria , Method according to one of the preceding claims, wherein the Kalman filtering is calculated in a real-time system. Method according to one of the preceding claims, wherein the multi-stage Kalman filtering for filtering signals in the electrical system of the means of transport (80) and in particular for filtering image signals and / or radar signals and / or ultrasound signals and / or LIDAR signals and / or GPS tracking Data of the means of transport (80) is used. Device for filtering a first signal (S) in the electrical system of a means of transport (80) comprising: A data input (12), • an evaluation unit (10), and A data output (14), wherein the evaluation unit (10) is set up, In conjunction with the data input (12) to filter the first signal (S) by means of a first Kalman filter (46) representing a first stage (42) for generating a second signal (S '), To filter the second signal (S ') by means of a second Kalman filter (48) representing a second stage (44) for generating a third signal (S "), and • Output the third signal (S ") via the data output (14). Means (80) comprising a device according to Claim 10 ,

Images