DE102019003679A1 - An engine analysis system for determining engine abnormality and methods for determining an engine abnormality - Google Patents

Abstract

The invention relates to an engine analysis system (10) for determining an engine abnormality (26b) of an engine (12) of a motor vehicle (14), comprising detection means (16) for detecting current vibrations (22) of the engine (12) and an electronic computing device (18) adapted to evaluate the detected vibrations (22), wherein the electronic calculation means (18) comprises at least one machine learning algorithm (20) and the evaluation of the current vibrations (22) is performed on the basis of past vibrations, the algorithm (20) is learned in response to past vibrations by a learning process for determining the engine abnormality (26b). Furthermore, the invention relates to a method.

Description

The invention relates to an engine analysis system for determining an engine abnormality of an engine of a motor vehicle, having a detection device for detecting current vibrations of the engine and having an electronic computing device, which is designed to evaluate the detected vibrations. Furthermore, the invention relates to a method for determining an engine abnormality of an engine of a motor vehicle.

Both in the production of a motor vehicle, as well as in the subsequent use of a motor vehicle, it can happen that unexpected errors, so-called engine anomalies, which are due to engine malfunctions, can occur. In such cases, it is currently necessary for trained and experienced personnel to use manual analysis to discover the cause of the error that has occurred. Due to the complexity of the operation and the complicated structure of a motor vehicle engine countless different causes of errors can lead to many more error symptoms, which often differ only by the smallest details and are therefore very difficult to identify and correct. This problem usually results in a very time-consuming and costly workshop stay.

The DE 11 2015 001 905 T5 discloses a method for determining a waveguide temperature for at least one waveguide used to generate a temperature map. The transceiver generates an acoustic signal passing through a measurement space defined by a wall, such as in a combustion chamber, in a hot gas flow path. The method includes calculating a total run time for the acoustic signal and subtracting a waveguide run time from the total run time to obtain a measurement runtime. A temperature map is calculated on the basis of the measuring room running time. An estimated wall temperature is obtained from the temperature map. Thereafter, an estimated waveguide temperature is calculated based on the estimated wall temperature, wherein the estimated waveguide temperature is determined without using a temperature sensing device.

The object of the present invention is to provide an engine analysis system and a method by means of which improved engine anomalies of an engine of the motor vehicle can be determined.

This object is achieved by an engine analysis system and by a method according to the independent patent claims. Advantageous embodiments are specified in the subclaims.

One aspect of the invention relates to an engine analysis system for determining an engine abnormality of an engine of a motor vehicle, having a detection device for detecting current vibrations of the engine and having an electronic computing device, which is designed to evaluate the detected vibrations.

It is envisaged that the electronic computing device has at least one machine learning algorithm and that the evaluation of the current vibrations based on past vibrations is carried out, wherein the algorithm is trained as a function of past vibrations by a learning process for determining the engine anomaly.

This can improve the engine abnormality can be determined. In particular, the algorithm is able to perform a much faster and more efficient root cause analysis on a damaged or incorrectly running engine. It is no longer necessary to use trained and experienced personnel for this work.

In particular, it is provided that the engine is designed as an electric motor or as an internal combustion engine.

According to an advantageous embodiment, the detection device is designed as a motor vehicle external or as a motor vehicle internal detection device.

Furthermore, it is provided in particular that the algorithm is trained by at least one classifier.

In a further advantageous embodiment, the algorithm is trained by a plurality of classifiers.

It has also been found to be advantageous if the engine analysis system is configured to determine the engine abnormality and an engine abnormality cause of the engine based on the actual vibrations.

A further aspect of the invention relates to a method for determining an engine anomaly of an engine of a motor vehicle by means of an engine analysis system, in which detected by a detection device, current vibrations of the engine and evaluated by means of an electronic computing device, the detected vibrations for determining the engine abnormality.

It is envisaged that by means of an algorithm for machine learning the electronic computing device based on past Vibrations are evaluated, the current vibrations, wherein the algorithm is learned in dependence of past vibrations through a learning process for determining the engine abnormality.

Advantageous embodiments of the engine analysis system are to be regarded as advantageous embodiments of the method. For this purpose, the engine analysis system has objective features which enable a performance of the method or an advantageous embodiment thereof. In particular, the method is carried out by means of the engine analysis system.

Further advantages, features and details of the invention will become apparent from the following description of a preferred embodiment and from the drawing. The features and feature combinations mentioned above in the description as well as the features and feature combinations mentioned below in the figure description and / or alone in the single figure can be used not only in the respectively indicated combination but also in other combinations or alone, without the frame to leave the invention.

The single FIGURE shows a schematic side view of an embodiment of a motor vehicle with an embodiment of an engine analysis system.

In the figure, identical or functionally identical elements are provided with the same reference numerals.

The FIGURE shows a schematic side view of an embodiment of an engine analysis system 10 , The engine analysis system 10 is for determining an engine abnormality 26b an engine 12 of a motor vehicle 14 educated. The engine analysis system 10 has a detection device 16 for detecting current vibrations 22 of the motor 12 on. In the present embodiment, the detection device 16 on the motor vehicle 14 educated. Alternatively or additionally, it may be provided that the detection device 16 also external to the motor vehicle, for example on a test stand for the motor vehicle 14 , is trained.

Furthermore, the engine analysis system 10 an electronic computing device 18 on which to evaluate the detected vibrations 22 is trained.

It is envisaged that the electronic computing device 18 an algorithm 20 for machine learning and evaluating the actual vibrations 22 based on past vibrations, the algorithm 20 in response to past vibrations through a learning process to determine the engine abnormality 26b is trained.

In particular, the Fig. Shows that by means of electronic computing device 18 an evaluation of the engine 24 can be carried out. Furthermore, the Fig. Shows that by means of electronic computing device 18 a motor anomaly cause 26a as well as the engine anomaly 26b based on the current vibrations 22 can be determined. It can thus be carried out an engine analysis.

The invention makes use of the idea that, in contrast to classical analysis methods which try to determine algorithms by means of limit values, partly by experience, partly by other analyzes, to make an assessment on the basis of machine learning. The algorithm 20 with machine learning, which can also be called an artificial intelligence algorithm, works exclusively with training data. It is not about whether the test object exceeds its limit, but rather tries the algorithm 20 to come to the right result. Here he tries to independently recognize a process or develops his own process, as he comes from the given input data to the corresponding result. In particular, a so-called supervised learning is used for this purpose. Alternatively, unsupervised learning can be used. The following is based on the Supervised Learning, in which both the input and the result for training the algorithm 20 or the electronic computing device 18 is given. Following this becomes a classifier 26 , which can also be referred to as a classifier, trained, which assigns the input data to each category. In particular, therefore, only the classifier becomes 26 trained.

In particular, it can be provided that the electronic computing device 18 to evaluate the current vibrations 22 has a neural network. The detection device 16 can be designed in particular as an acoustic detection device and so the corresponding vibrations 22 take up. Alternatively or additionally, the detection device 16 based on, for example, a force detection the vibrations 22 of the motor 12 take up.

As training data for the algorithm described above 20 serve at the beginning of the provided by the external sensors vibration data of the engine 12 , These are prepared in particular by a preprocessing and later as training data in the classifier 26 given. This training record includes recorded Data snippets of, for example, five seconds in length with the appropriate identifier, whether the engine 12 is okay, which is shown in the figure by a hook, or whether it is not in order, which is shown in the figure as a cross or the corresponding troubleshooting.

Training will then be a combination of different classifiers 26 used to exploit the respective strengths of the individual and thus increase the overall accuracy. By training with large amounts of data, it is the classifier 26 so possible to keep improving.

In other words, it can be provided that the algorithm 20 by at least one classifier 26 is trained. Alternatively, the algorithm may be implemented by a variety of classifiers 26 to be trained.

From the training phase then goes a fully trained classifier 26 which is optimized to still identify and return the correct identifier for new and thus unknown input data. As the training is complete, the classifier becomes 26 no further training and thus you get to exactly the same input sequence and the exact same identifier. Thus, the method can additionally detect defective motors 12 also the cause of the error, in other words the engine anomaly cause 26a , determine and thus supportive for testing of the engine 12 be used.

Overall, the invention shows a detection of engine anomalies 26b for fault cause analysis by artificial intelligence.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 112015001905 T5 [0003]

Claims (6)

An engine analysis system (10) for determining an engine abnormality (26b) of an engine (12) of a motor vehicle (14), comprising detection means (16) for detecting current vibrations (22) of the engine (12) and electronic computing means (18), which is designed for evaluating the detected vibrations (22), characterized in that the electronic calculation device (18) has at least one machine learning algorithm (20) and the evaluation of the current vibrations (22) is carried out on the basis of past vibrations, wherein the Algorithm (20) is learned in response to the past vibrations by a learning process for determining the motor anomaly (26b). Motor analysis system (10) according to Claim 1 , characterized in that the detection device (16) is designed as a motor vehicle external or as a motor vehicle internal detection device. Motor analysis system (10) according to one of Claims 1 or 2 , characterized in that the algorithm (20) is taught by at least one classifier (26). Motor analysis system (10) according to one of the preceding claims, characterized in that the algorithm (20) is trained by a plurality of classifiers (26). An engine analysis system (10) according to any one of the preceding claims, characterized in that the engine analysis system (10) is adapted to determine an engine abnormality (26b) and an engine abnormality cause (26a) of the engine (12) based on the actual vibrations (22). Method for determining an engine anomaly (26a) of an engine (12) of a motor vehicle (14) by means of an engine analysis system (10), in which current vibrations (22) of the engine (12) are detected by a detection device (16) and detected by an electronic computer ( 18) the detected vibrations (22) are evaluated for determining the engine anomaly (26b), characterized in that the current vibrations (22) are evaluated by means of an algorithm (20) for machine learning of the electronic computing device (18) on the basis of past vibrations, wherein the algorithm (20) is learned in response to the past vibrations by a learning process for determining the engine abnormality (26b).

Images