Method for detecting anomalies on an echo measuring device, method for providing an artificial intelligence, computer program, and computer-readable medium

Abstract

The invention relates to a method for detecting anomalies on an echo measuring device (1), said echo measuring device (1) comprising at least one receiving unit (3) for receiving an echo signal, wherein the method has the following steps: a) receiving an echo signal by means of the receiving unit (3) and analyzing the echo signal; b) detecting anomalies in the echo signal using at least one trained artificial intelligence; and c) outputting information indicating the presence of an anomaly if an anomaly is detected.

Description

[0001] Description

[0002] Method for detecting anomalies in an echo measuring device and method for providing artificial intelligence, as well as computer program and computer-readable medium

[0003] The present invention relates to a method for detecting anomalies in an echo measuring device, wherein the echo measuring device has at least one receiving unit for receiving an echo signal. Furthermore, the present invention relates to a method for providing a trained artificial intelligence for detecting anomalies in an echo measuring device, wherein the echo measuring device has at least one receiving unit for receiving an echo signal.

[0004] Echometers are used in a variety of applications. For example, radar or ultrasonic devices can be used to measure distances between the echometer and a surface or to record a surface profile even over long distances. Such devices can be based on time-of-flight measurement, i.e., a signal is emitted by the device, and the time until a reflected or backscattered signal reaches a receiving unit of the echometer is determined.

[0005] As a result, the echo measuring device usually outputs a measured value, e.g. the distance between the echo measuring device and a surface, a fill level or a surface profile. The measured value is determined from an echo profile. Such an echo profile includes, among other things, the reflections of objects in the area covered by an emitted signal. However, the echo profile is also influenced by other factors. For example, there may be dirt on an antenna, which may be part of a receiving unit, which disrupts the echo profile. Other causes of an incorrect echo profile can be incorrect installation. For example, such an echo measuring device may be mounted at an incorrect angle to the surface to be detected. Interfering objects may also be located in the area of ​​the usually club-shaped emitted signal.Multiple reflections can also occur, affecting the echo profile. In this case, the first signal in the echo profile is the "true" signal, and additional signals with multiple distances or travel times of the true echo are then multiple reflections.

[0006] In other words, anomalies in an echo profile can be caused by faulty installation or interfering objects in the detected area, and / or by multiple reflections that ultimately result from the same detected object. This allows faulty installations, wear, or deposits on the echo measuring device to be detected. In addition, interfering objects can be identified, and the wanted signal can be differentiated from other interference signals to reliably identify the wanted signal in the echo profile.

[0007] Typically, an experienced echometer user can determine whether the installation is correct. This is usually based on their experience from numerous installations and corrections of echometer installations. Furthermore, an experienced user can also determine whether multiple reflections are occurring.

[0008] Although experienced users are able to detect an anomaly in an echo signal based on their experience, it can be considered a disadvantage that such experienced users are not always available or have only a limited working capacity.

[0009] The object of the present invention is therefore to create a method for detecting anomalies in an echo measuring device, which is less personnel-intensive and less dependent on the personal experience of a user. This object is achieved in a method of the type mentioned above.

[0010] This object is achieved by comprising the following steps: a) receiving an echo signal by the receiving unit and evaluating this echo signal; b) detecting anomalies in the echo signal using at least one trained artificial intelligence; c) outputting information that an anomaly is present when an anomaly is detected.

[0011] The method is carried out on an echo measuring device which comprises at least one transmitting unit for emitting a signal and at least one receiving unit for receiving an echo signal. The echo signal is caused by the signal emitted by the transmitting unit, for example by a reflection of the signal. The invention is based on the fundamental idea of ​​using artificial intelligence to carry out a quasi-automated evaluation of the echo signal. Such trained artificial intelligence is able to recognise faulty installations as deviations from usual echo signals on the basis of its training and thus to distinguish good installations from bad ones. The echo signal can have the form of an echo profile. This means that the intensity of a signal is displayed or determined over the propagation time. In particular, several echo signals can be recorded one after the other.In particular, the entire echo signal can be evaluated directly. In the method according to the invention, the echo measuring device is designed to measure a level in an industrial application. Such an industrial application can be designed, for example, as a tank or silo. The echo measuring device can be attached in or to the industrial application. The level can, for example, be the surface of a liquid or a bulk material. To determine the anomalies, reference is made not only to the current echo profile but also to previous profiles. In this case, for example, the speed of a peak, i.e. the change in the position of a peak or the amplitude of a peak over time, or an amplitude change and / or the minimum / maximum position of a peak over time in the echo profiles can be used.

[0012] The method may further comprise a classification of the anomaly, in particular the output of a cause of the anomaly. The classification may be performed, for example, using artificial intelligence. In other words, the method may not be limited to detecting that an anomaly exists; rather, the method may also determine a cause of the anomaly based on at least one trained artificial intelligence.

[0013] Anomalies can, on the one hand, be deviations in the echo profile resulting from faulty installation. It is also possible that the anomaly is based on multiple reflections, and thus the method according to the invention can be characterized by the selection of the correct echo from a large number of echoes in an echo profile. For example, in the case of multiple reflections, a large number of echoes occur that are separated by a multiple of the propagation time of a first echo, with the intensity of the subsequent echoes decreasing with increasing propagation time.

[0014] The method according to the invention can further comprise the output of a remedial measure for eliminating the anomaly. The remedial measure is determined using the artificial intelligence, and the determined remedial measure is output by the artificial intelligence. The output can be made directly on the echo measuring device or on an external computer, in particular on a human-machine interface, preferably on a display. Preferably, one artificial intelligence is used jointly for the evaluation of the entire echo profile. This is understood to mean, for example, the entire echo profile caused by the associated signal emitted by the transmitting unit. It is also conceivable that different trained artificial intelligences are used for different parts of the echo profile. The artificial intelligence can have been trained at least partially with simulated measurement data.The simulated measurement data can be generated, for example, by a simulator, which can be based on physical models of the echo measuring device. Such a simulator can be designed as a digital twin of the echo measuring device.

[0015] In a further embodiment, the echo measuring device can perform a transit time measurement. The echo measuring device can be an ultrasonic device or a radar device. These are common devices designed to emit signals and detect or record an echo of these signals.

[0016] The echo signal or information from an echo signal or data relating to an echo signal or echo profile can be sent to a central computer unit and / or to a cloud so that possible anomalies can be detected there. In other words, in this case, no local detection of anomalies takes place in the echo measuring device itself, but rather in an external computer unit or in a cloud. This has the advantage that a large amount of data can be stored centrally there, which the artificial intelligence can access. At the same time, the data from various echo measuring devices can be collected there and used for further training of the artificial intelligence.

[0017] Alternatively, an anomaly can be detected locally in the echo measuring device. In this case, no complex data transmission to an external computer unit is required. According to a preferred embodiment, the method further comprises extracting a useful signal from the echo signal. In other words, it can be provided that, starting from the echo signal, in particular an echo profile, a relevant useful signal is determined by eliminating detected interference signals from the echo signal. The method can also comprise outputting a determined useful signal or useful echo.

[0018] The extraction of the useful signal, for example a distance indication, can not only be carried out on the basis of the existing or current echo signal. The method or the trained artificial intelligence can also be designed in such a way that a useful signal is determined from previous echo signals or with the aid of previous echo signals. For this purpose, for example, changes in the amplitude of local maxima or peaks in the echo signal in the form of an echo profile, the movement speed of local maxima or the minimum and maximum position of local maxima or peaks can be used. In other words, several echo signals can be received one after the other in which changes occur over time, from which information about anomalies or interference signals can be obtained.Such an approach is based on the idea that, with echo signals that vary over time, certain signals, particularly peaks or maxima, remain in their position, while certain useful signals have a variable position. It is also possible that moving parts repeatedly lead to the same echo signals. Thus, to determine the position of a useful signal, not only the current echo signal, for example in the form of an echo profile, can be used, but also previous echo signals or echo profiles.

[0019] The object underlying the invention is further achieved by a method for providing a trained artificial intelligence for detecting anomalies in an echo measuring device of the type mentioned above, which method comprises the following steps:

[0020] - receiving input training data representing an echo signal received by the receiving unit of an echo measuring device, in particular the complete echo signal;

[0021] - receiving output training data representing anomalies in the echo signals, the output training data comprising mappings to the input training data;

[0022] - Training the artificial intelligence based on the input training data and the output training data in such a way that these anomalies are detected simultaneously;

[0023] - Providing the trained artificial intelligence.

[0024] This design is based on the idea that the artificial intelligence is trained by receiving input training data which is based on an echo signal received by a receiving unit. In particular, this can be the complete echo signal or echo profile. Depending on what certain training data looks like, certain output training data can be assigned to it. For example, assignments can be made to certain anomalies, such as faulty installation or multiple reflections. The artificial intelligence is then trained so that it is able to perceive different anomalies simultaneously. This is because echo signals often contain not just one anomaly, but several or even a large number of anomalies at the same time.In other words, this makes it possible to reliably detect various anomalies from an echo signal. The echo measuring device is designed to measure a level in an industrial application. The output training data can further include a classification of anomalies and / or an output of a remedial measure to eliminate the respective anomaly.

[0025] A computer program (or computer program product) according to the invention comprises instructions which, when the program is executed on a computer, cause the computer to carry out methods according to the invention for detecting anomalies.

[0026] A further computer program (or computer program product) according to the invention comprises instructions which, when the program is executed on a computer, cause the computer to carry out the method described above for providing a trained artificial intelligence.

[0027] Furthermore, the present invention comprises a computer-readable medium comprising a computer program as described above.

[0028] For further details of the invention, reference is made to the dependent claims and the description of the following embodiment with reference to the drawing. The drawing shows:

[0029] Figure 1 is a schematic representation of a method according to the invention for detecting anomalies on an echo measuring device.

[0030] In Figure 1, an echo measuring device 1 is shown at the top, which has a transmitting unit 2 for emitting a signal and a receiving unit 3 for receiving an echo signal. The signals emitted by the transmitting unit 2 are reflected on the schematically shown surface 4, which can be a level in an industrial application, for example a liquid or a bulk material. Such an industrial application can be, for example, a tank or a silo, in or on which the echo measuring device is attached. Specifically, the echo measuring device 1, which is designed, for example, as a radar device, carries out a propagation time measurement, i.e. the time offset between the signal emitted by the transmitting unit 2 and the echo signal detected by the receiving unit 3 is evaluated. An echo profile is shown schematically in Figure 1 below the arrangement.The propagation time t is indicated on the horizontal axis, and the intensity I of the echo signal is generally indicated on the vertical axis. The entire echo signal is preferably evaluated. The echo signal or echo profile can be transmitted to an external computer unit 5. There, an evaluation and, in particular, a detection of anomalies can take place using a trained artificial intelligence 6. As soon as an anomaly is detected, information that an anomaly exists can be output either via the external computer unit 5 or directly on the echo measuring device 1. Furthermore, the artificial intelligence can classify the anomaly so that the cause of the anomaly can also be output. It is also possible for a remedial measure to be output to eliminate this anomaly.The output of the information that an anomaly exists can, for example, take place on a display 7 which is connected to the external computer unit 5.

[0031] Although the invention has been illustrated and described in detail by the preferred embodiment, the invention is not limited to the disclosed examples and other variations can be derived therefrom by those skilled in the art without departing from the scope of the invention.

[0032] Regardless of the grammatical gender of a particular term, persons with male, female or other gender identity are included.

Claims

Patent claims 1. A method for detecting anomalies in an echo measuring device (1), wherein the echo measuring device (1) has at least one transmitting unit (2) for emitting a signal and at least one receiving unit (3) for receiving an echo signal, wherein the echo signal is caused by the signal emitted by the transmitting unit (2), and wherein the method comprises the following steps: a) receiving the echo signal by the receiving unit (3) and evaluating this echo signal; b) detecting anomalies in the echo signal using at least one trained artificial intelligence (6); c) outputting information that an anomaly is present if an anomaly is detected;characterized in that for evaluating the echo signal and / or for detecting anomalies in the echo signal, reference is made to previous echo signals by means of the trained artificial intelligence (6), wherein at least the speed of a peak in the echo signal, the amplitude change of a peak and / or the minimum and / or maximum position of a peak in an echo profile is evaluated or used over time, and the echo measuring device (1) is designed to measure a level in an industrial application.; 2. Method according to claim 1, characterized in that it further comprises a classification of the anomaly, in particular the output of a cause of the anomaly.

3. Method according to claim 1 or 2, characterized in that it further comprises determining and issuing a remedial measure for eliminating the anomaly by means of the artificial intelligence (6).

4. Method according to one of the preceding claims, characterized in that an artificial intelligence (6) is used to evaluate the entire echo profile caused by the emitted signal.

5. Method according to one of the preceding claims, characterized in that the artificial intelligence (6) is / was trained at least partially with simulated measurement data generated with a simulator based on physical models of the echo measuring device (1).

6. Method according to one of the preceding claims, characterized in that the echo measuring device (1) carries out a transit time measurement and / or is an ultrasound device or a radar device or a ladar device.

7. Method according to one of the preceding claims, characterized in that the echo profile is sent to a central computer unit and / or a cloud and possible anomalies are detected there.

8. Method according to one of claims 1 to 6, characterized in that the anomalies are detected locally in the echo measuring device (1).

9. Method according to one of the preceding claims, characterized in that the method further comprises extracting a useful signal from the echo signal.

10. A method for providing a trained artificial intelligence for detecting anomalies on an echo measuring device (1), wherein the echo measuring device (1) has at least one receiving unit (3) for receiving an echo signal, comprising the following steps: - receiving input training data representing an echo signal received by the receiving unit (3); - receiving output training data representing anomalies in the echo signals, the output training data comprising mappings to the input training data; - Training the artificial intelligence based on the input training data and the output training data in such a way that several different anomalies can be detected simultaneously in an echo signal; - Providing the trained artificial intelligence; wherein the echo measuring device (1) is designed to measure a level in an industrial application.

11. Method according to claim 10, characterized in that the output training data further comprises a classification of anomalies and / or an output of a remedial action for eliminating the anomalies.

12. A computer program comprising instructions which, when the program is executed on a computer, cause the computer to carry out the method according to any one of claims 1 to 11.

13. A computer program comprising instructions which, when executed on a computer, cause the computer to carry out the method of claim 10 or 11.

14. A computer-readable medium comprising a computer program according to claim 12 or 13.

Images