Method and device for improved evaluation of measurement signals from a sensor

Abstract

Method for evaluating a measurement signal (M) of a sensor which is configured to determine at least one variable (A) characterizing an operating state of a technical system, wherein the evaluation is carried out by means of a model (301) and a trained machine learning system (302), wherein a weighting of the model (301) and the machine learning system (302) is carried out depending on how similar the measurement signal (M) is to training data on the basis of which the machine learning system (302) was trained.

Description

The invention relates to a method for evaluating a measurement signal of a sensor, a computer program, a machine-readable storage medium and a control device.Prior ArtAccording to the prior art, a virtual sensor can be used to determine a target variable which depends on one or more correlating measured variables. In contrast to conventional physical sensors, the target variable is not measured directly, but rather is mapped by software on the basis of correlations with other measured variables. Mathematical models, simulations or artificial intelligence can be used here.For example, DE 10 2023 206 032.9, which is not prepublished, discloses a method for evaluating a measurement signal of a sensor which is configured to determine at least one variable characterizing an operating state of a technical system, wherein a low-frequency component of the measurement signal is determined for this purpose and / or a high-frequency component of the measurement signal is determined, and wherein the evaluation of the low-frequency component and the evaluation of the high-frequency component are carried out using different evaluation methods for determining the variable characterizing the operating state of the technical system.Advantages of the InventionThe invention with the features of independent claim 1 has the advantage over the related art that it delivers particularly reliable results in wide ranges of measurement signals.Further aspects of the invention are the subject of the subordinate claims. Advantageous further developments are the subject of the dependent claims.Disclosure of the InventionIn a first aspect, the invention relates to a method for evaluating a variable characterizing a measurement signal of a sensor, wherein the evaluation is carried out by means of a model, in particular a physical model, and a trained machine learning system, wherein the model and the machine learning system are weighted depending on how similar the measurement signal is to training data, on the basis of which the machine learning system has been trained.In other words, this method is a virtual sensor for the variable characterizing the operating state of the technical system.The similarity between the measurement signal and the training data is a particularly simple measure for ascertaining a reliability of the machine learning system. It is thus possible to determine the variable characterizing the operating state of the technical system as reliably as possible.In some embodiments, it is provided that the measurement signal is provided to the model and the trained machine learning system at a respective input of the model and the trained machine learning system.In some embodiments, it is provided that an estimated value of the variable characterizing the operating state of the technical system is provided at an output of the model, and that a correction value of this estimated value is provided at an output of the trained machine learning system.In other embodiments, it can be provided that an estimated value of the variable characterizing the operating state of the technical system is provided in each case at outputs of the model and of the machine learning system.In further developments, it can be provided that the similarity of the measurement signal to the training data is carried out by means of an out-of-distribution detection method.The term "out-of-distribution detection" derives from English and refers to methods for detecting whether the measurement signal originates from the same statistical distribution as the training data.Furthermore, it can be provided that the similarity of the measurement signal to the training data is determined with the aid of an autoencoder trained with the training data.This is a particularly simple way of performing out-of-distribution detection. In other words, in some embodiments, autoencoder and machine learning system are trained with the same training data.In some embodiments, it can be provided that the similarity is determined as a function of a reconstruction error if the measurement signal is applied to an input of the autoencoder and a reconstruction of this measurement signal is provided at an output of the autoencoder.In some embodiments, it may be provided that the contributions of the machine learning system and of the model are combined additively and / or multiplicatively.These combination possibilities are particularly simple ways of combining both contributions of the model and contributions learned on the basis of data.In some embodiments, it may be provided that the weighting evaluates a contribution of the machine learning model to be lower the lower the similarity between measurement signal and training data.A "smaller contribution" is to be understood here in some embodiments such that a correction of an output of the model effected by the machine learning system is smaller.In other words, this level of the contribution causes weighting as to how the ascertained value of the virtual sensor is composed of the machine learning system and the model. In some embodiments, the autoencoder reconstruction error determines this weight. In this case, the proportion of the model in the ascertained value of the virtual sensor increases with increasing reconstruction error.This makes it possible to determine the variable characterizing the operating state of the technical system as a determined value of the virtual sensor as reliably as possible.Embodiments of the invention are explained in more detail below with reference to the attached drawings. In the drawings, there are shown: FIG. 1 schematically shows a sensor in a brake system; FIG. 2 schematically shows an exemplary embodiment of a device for ascertaining a variable which characterizes the operating state of the technical system; FIG. 3 schematically shows a further exemplary embodiment of a device for determining a variable which characterizes the operating state of the technical system; FIG. 4 schematically shows an apparatus for training an autoencoder and one of the aforementioned apparatuses; FIG. 5 schematically shows a flow chart of a sequence according to an embodiment of the invention.FIG. 1 schematically shows a brake system (1) of a motor vehicle (11) having, in the exemplary embodiment, a plurality of sensors (2) comprising a pressure sensor in a hydraulic feed line (3), with which a contact pressure force of a brake caliper (4) is controlled, a sensor for determining a longitudinal acceleration of the motor vehicle (11), a sensor for detecting a temperature of the liquid in the hydraulic feed line (3) and a voltage sensor on a pump (not shown) for conveying the liquid. A multidimensional measurement signal (M) comprising the measurement signals of these sensors is supplied to a control unit (10), which comprises a computer-readable storage medium (20), on which a computer program is stored, on which a program for evaluating the measurement signal (M) is stored in order to determine a torque that the brake caliper (4) transmits to a wheel (not shown) of the motor vehicle (11). The control device (10) further comprises a processor (21) which is configured to execute computer programs stored on the storage medium (20).FIG. 2 schematically shows an embodiment of a device ( 300) for evaluating the measurement signal (M) for ascertaining the variable (A) characterizing the operating state of the technical system-in the exemplary embodiment the torque that the brake caliper ( 4) transmits to the wheel. In the exemplary embodiment, this device ( 300) is stored as a computer program on the storage medium ( 20) and is executed by the control unit ( 10).This device (300) comprises a physical model (301) comprising a physical modeling of the relationship between the measurement signal (M) and the transmitted torque (A). The measurement signal (M) is supplied to an input (N 1) of the physical model ( 301), which provides an estimated value (AP) of the transmitted torque at an output (X 1) of the physical model ( 301) by means of the physically modeled relationship.The device ( 300) further comprises a machine learning system ( 302, in the exemplary embodiment a neural network. The measurement signal (M) is fed to the neural network ( 302) at a first input (N 2) and the estimated value (AP) of the transmitted torque is fed to a second input (N 3). The neural network ( 302) determines from this a correction value (DA) which is provided at an output (X 2) of the neural network ( 302).Estimated value (AP) and correction value (DA) are fed to a weighted combiner ( 303), which in the exemplary embodiment determines from a weighted sum of estimated value (AP) and correction value (DA) the torque (A) that the brake caliper ( 4) is transmitting to the wheel.In embodiments, this is done using the formula with a predefinable weighting factor (b).The apparatus (300) further comprises a weighting block (250) which specifies the weighting factor (b) as a function of the measurement signal (M). In the exemplary embodiment, the weighting block ( 250) comprises an autoencoder (German: "autoencoder") which is configured to determine a reconstruction of the measurement signal (M). The weighting block ( 250) is configured to ascertain the weighting factor (b) as a function of the reconstruction error between measurement signal (M) and reconstruction of the measurement signal (M). In the exemplary embodiment, a characteristic curve is stored in the weighting block ( 250), which curve specifies the value b=1 in the case of reconstruction errors equal to zero and decreases with increasing reconstruction errors, where b→0 in the case of large reconstruction errors.FIG. 3 shows in an exemplary embodiment a further embodiment of the device ( 300) for evaluating the measurement signal (M) for ascertaining the variable (A) characterizing the operating state of the technical system-in the exemplary embodiment the torque that the brake caliper ( 4) transmits to the wheel. In the exemplary embodiment, this device ( 300) is stored as a computer program on the storage medium ( 20) and is executed by the control unit ( 10).This device ( 300) also comprises a physical model ( 301) which comprises a physical modeling of the relationship between measurement signal (M) and the transmitted torque (A). The measurement signal (M) is supplied to an input (N 1) of the physical model ( 301), which provides an estimated value (AP) of the transmitted torque at an output (X 1) of the physical model ( 301) by means of the physically modeled relationship.The device ( 300) further comprises a machine learning system ( 302, in the exemplary embodiment a neural network. The measurement signal (M) is fed to the neural network ( 302) at a first input (N 2). The neural network ( 302) determines an estimated value (MA) of the transmitted torque from this value, which is provided at an output (X 2) of the neural network ( 302).The two estimated values (AP, AM) are fed (DA) to a weighted combiner ( 303), which in the exemplary embodiment determines the torque (A) from a weighted sum of the estimated values (AP, AM) that the brake caliper ( 4) is transmitting to the wheel.In embodiments, this is done using the formula with predefinable weighting factors (a), (b). These are preferably set under the constraint.The apparatus (300) further comprises a weighting block (250) which specifies the weighting factors (a), (b) as a function of the measurement signal (M). In the exemplary embodiment, the weighting block ( 250) comprises an autoencoder (German: "autoencoder") which is configured to determine a reconstruction of the measurement signal (M). The weighting block ( 250) is configured to ascertain the weighting factors (a), (b) as a function of the reconstruction error between measurement signal (M) and reconstruction of the measurement signal (M).The determination of the weighting factor (b) takes place in the exemplary embodiment analogously as discussed in connection with FIG. 2. The weighting factor (a) is determined by means of.FIG. 4 shows, in an exemplary embodiment, a mode of operation of a training device ( 100) for training the devices ( 300) illustrated in FIGS. 2 and 3 for ascertaining the variable (A) characterizing the operating state of the technical system as a function of the measurement signal (M).The apparatus (100) comprises a provision block (105) for providing measurement signals (M) from a training dataset comprising a plurality of measurement signals (M) and associated setpoint values of variables (AS) characterizing the operating state of the technical system.The training device (300) comprises a processor (120) for executing the training algorithm realized as a computer program, which is stored on a machine-readable storage medium (121) of the training device (300).The device ( 300) is supplied with the measurement signals (M), which determine from this the variable (A) characterizing the operating state of the technical system. The associated setpoint variable (A') is fed to a comparison device (110), which, depending on a cost function dependent on this variable (A) and associated setpoint variable (A'), adjusts parameters of the apparatus (300) and, in particular, of the machine learning system (302) in a manner usually as monitored training, such that the apparatus (300) is able to output the variable (A) correctly as well as possible.The training device ( 100) is likewise configured to train the autoencoder ( 200) unsupervisedally with the measurement signals (M) of the same training dataset. The autoencoder (200) comprises, in the usual manner, an encoder (201) which converts the measurement signals (M) into a latent vector (L) of lower dimensionality, and comprises a decoder (201) which determines a reconstruction (M') of the measurement signal (M) from this latent vector (L). The measurement signals (M) are transmitted to the comparison device (110). From a comparison of measurement signal (M) and associated reconstruction (M'), the parameters of the autoencoder (200) are adjusted in the usual manner by means of a cost function which penalizes the distance between measurement signal (M) and reconstruction (M') in such a way that the reconstruction (M') approximates the associated measurement signal (M) as well as possible.FIG. 5 shows a flow chart of an embodiment of the method as it can run in the control unit ( 10). In some embodiments, not belonging to the method, a previous training ( 1000) of the device ( 300) and the autoencoder ( 300) by means of the training device ( 100).Subsequently, during operation, measurement data (M) are supplied to the device ( 300). (1100). Here, weighting parameters (a), (b) are determined by means of the weighting block (250) and fed to the combiner (303).As illustrated in Figures (2) and (3), physical model (301) and machine learning model (302) determine their respective contributions (AP), (AM), (DA). And feed them to the combiner (303) (1200).In combiner ( 303), these contributions are combined according to formula ( 1) or formula ( 2) in order to determine ( 1300) variable (A) characterizing the operating state of the technical system. The method ends with this.References included in the specificationThis list of documents cited by the applicant has been produced in an automated manner and is only included for the better information of the reader. The list is not part of the German patent application or utility model application. The DPMA does not take any adhesion for any faults or omissions.Patent Literature citedDE 10 2023 206 032.9

[0003]

Claims

Method for evaluating a measurement signal (M) of a sensor (2) which is configured to determine at least one variable (A) characterizing an operating state of a technical system (11), wherein the evaluation is carried out by means of a model (301) and a trained machine learning system (302), wherein the model (301) and the machine learning system (302) are weighted on the basis of how similar the measurement signal (M) is to training data on the basis of which the machine learning system (302) has been trained.Method according to Claim 1, wherein the similarity of the measurement signal (M) to the training data is carried out by means of an out-of-distribution detection method.Method according to Claim 2, wherein the similarity of the measurement signal (M) to the training data is determined with the aid of an autoencoder (200) trained with the training data.Method according to Claim 3, wherein the similarity is determined as a function of a reconstruction error if the measurement signal (M) is applied to an input of the autoencoder (200) and a reconstruction of this measurement signal (M') is provided at an output of the autoencoder.Method according to one of the preceding claims, wherein the contributions of the machine learning system (AM, AD) and of the model (AP) are combined additively and / or multiplicatively.Method according to one of the preceding claims, wherein the weighting (b) evaluates a contribution of the machine learning model (AM, AD) to be lower the lower the similarity between measurement signal (M) and training data.A computer program adapted to cause a computer (10) to perform the method of any one of claims 1 to 6 when executed on the computer (10).A machine readable storage medium (21) on which the computer program of claim 7 is stored.Control device (10) which is configured to carry out the method according to one of Claims 1 to 6.

Images