DE102018210520B3 - Method and system for diagnosing a machine process - Google Patents

Abstract

A method according to the invention for classifying input data (X _D ) of a machine process, in particular of a machine (1), comprises the steps:
- generating (S20) notional input data (F) of a first machine process having a second state based on acquired first input data (X ₁ ) of said machine process having a first state by means of machine learning;
- training (S30) a classification method (K) on the basis of the fictitious and first input data; and
Classifying (S50) diagnostic input data (X _D ) of the first and / or at least one second machine process using the trained classification method.

Description

The present invention relates to a method and system for classifying input data of a machine process, a method and system for diagnosing a machine process by means of such a classification and a computer program product for carrying out a method described here.

The classification of input data of a machine process, for example a working robot or the like, can improve a diagnosis of the process, for example of the robot, in particular its monitoring, quality assurance and / or predictive maintanance.

So is out of the DE 44 47 288 A1 An engine failure diagnostic system and method is known in which machine conditions are diagnosed using a neural network.

The object of the present invention is to improve a classification of input data of a machine process, in particular a diagnosis of a machine process.

This object is achieved by a method having the features of claims 1 and 5, respectively. Claims 7, 8 protect a system or computer program product for carrying out a method described herein. The subclaims relate to advantageous developments.

• - Generieren von fiktiven Eingangsdaten eines ersten Maschinenprozesses, insbesondere wenigstens einer ersten Maschine, in einer Ausführung wenigstens eines ersten Roboters, der bzw. die einen zweiten Zustand aufweist, auf Basis erfasster erster Eingangsdaten dieses Maschinenprozesses bzw. dieser Maschine, der bzw. die einen ersten Zustand aufweist, mittels maschinellem Lernen;
• - Trainieren eines Klassifikationsverfahrens auf Basis der (generierten) fiktiven und (erfassten) ersten Eingangsdaten; und
• - Klassifizieren von Diagnose-Eingangsdaten des ersten und/oder wenigstens eines, insbesondere typgleichen oder typunterschiedlichen, zweiten Maschinenprozesses, insbesondere der ersten und/oder wenigstens einer zweiten Maschine, insbesondere gleichen oder unterschiedlichen Typs, in einer Ausführung des ersten und/oder wenigstens eines, insbesondere typgleichen oder typunterschiedlichen, zweiten Roboters, mithilfe des trainierten Klassifikationsverfahrens.

According to one embodiment of the present invention, a method for classifying one- or multi-dimensional input data of a machine process, in particular of at least one machine, in an embodiment of at least one robot, comprises the steps:

• Generating fictitious input data of a first machine process, in particular of at least one first machine, in an embodiment of at least one first robot having a second state based on detected first input data of that machine process or machine, the first one State, by means of machine learning;
• - training a classification method on the basis of the (generated) fictitious and (recorded) first input data; and
• Classifying diagnostic input data of the first and / or at least one, in particular type-identical or type-different, second machine process, in particular of the first and / or at least one second machine, in particular of the same or different type, in one embodiment of the first and / or at least one in particular type-identical or type-different, second robot, using the trained classification method.

In one embodiment, a trained classification method can improve the classification of diagnostic input data, in particular faster, more reliable and / or more precise.

By generating fictitious additional input data for a second state of a machine process, in turn, the training of the classification method can be improved, in particular the classification method trained on the basis of this additional (generated fictitious input data) can classify diagnostic input data faster, more reliably and / or more precisely.

In machine processes, as a rule, many input data are available for a fault-free state of the machine process, but only a few input data for a faulty state, since the machine process has one (if at all) rarely.

This complicates the (successful) training of classification methods, since a larger number (also) of negative examples, preferably an approximately equal number of positive and negative examples, is often advantageous or necessary.

Thus, in one embodiment, by generating fictitious input data for a second state of a machine process, training of a classification method based on acquired input data for a healthy first state and the additional fictitious input data for a second dirty state can be improved.

Accordingly, in one embodiment, the first state is a faultless or good or I (n) O (rdnung) state of the (respective) machine process, in particular one or more of its components, in particular machines, and the second state is a faulty or Poor or N (not) I (n) O (rdnung) state of this machine process or its component (s) or machine (s), without the present invention being limited thereto.

A machine process in the sense of the present invention may in one embodiment comprise, in one embodiment, an industrial, working process of one or more machines, in particular robots, in particular such.

Correspondingly, input data, in particular (fictitious) input data (acquired) first input data, diagnostic input data (to be classified or classified) and / or subsequently mentioned (acquired) second input data, operating parameters (values) of one or more machines, in particular one or more robots, in particular.

In one embodiment, these operating parameters may include vibration data, temperature data, force data and / or current or voltage data of the machine (s), in particular, or of vibrations of the machine (s), temperatures of the machine (s), in particular of drives, Driven, fluids of the machine (s) or the like, forces, in particular on drives, gearboxes and / or tools of the machine (s) or the like, and / or current or voltage data of the machine (s), in particular of drives of the machine ( n) or the like.

With such input data, machine processes, in particular states, can be evaluated and / or predicted particularly advantageously in one embodiment, in particular quickly (er), reliably (er) and / or precisely (r).

In one embodiment, if diagnostic input data of that machine process is classified (using a trained classification method), with the first input data of which the fictitious input data are also generated and / or the classification method has been trained, these diagnostic input data can be particularly advantageous in one embodiment. in particular, quickly (er), reliably (er) and / or accurately (r) assessed and / or predicted.

In another embodiment, if diagnostic input data of a machine process is classified using a trained classification method, which in turn is trained with first input data of another machine process and fictitious input data that has been generated based on first input data of the other machine process In one embodiment, diagnostic input data may also be classified by machine processes, difficult or impossible to obtain for the first input data for training purposes, and / or poor in quality and / or quantity, for example, noisy and / or only in small numbers.

In one embodiment, the classification method may include, in particular, a classification method or a classifier of machine learning, in particular supervised learning. In one embodiment, it maps (diagnostic) input data to output data or classes or assigns (diagnostic) input data to output data or classes, wherein the mapping or assignment in one embodiment is based on (diagnostic) input data and predetermined therefor Output data or classes is adapted or trained in order to achieve the best possible match between the predetermined and determined by the classification process output data or classes, or is set up for this purpose or is used for this purpose.

In one embodiment, the notional input data is generated using a Generative Adversary Network (GAN) network. Such GANs are described, for example, in Goodfellow, Yoshua Bengio, et al .: Deep Learning (Adaptive Computation and Machine Learning), MIT Press, Cambridge (USA), 2016. ISBN 978-0262035613.

In one embodiment, the two artificial neural subnetworks perform a zero-sum game, wherein one subnetwork or generator creates candidates according to a generative model and the other neural subnetwork or a discriminator evaluates the candidates according to a discriminating model and the one subnetwork or the generator tries to give the other neural subnetwork or the discriminator a bad answer or to generate as realistic values as possible, while the other neural subnetwork or the discriminator learns to avoid "poorly" defined results or tries to find out if it is is a real or a faked data point. In one embodiment, a so-called conditional GAN is used, in which the generator in one embodiment is first conditioned via a data point from the source domain so that it can finally make a style transfer.

As a result, the generation of fictitious, additional input data can be improved, in particular fast (er), reliable (er) and / or precise (r) can be performed.

• - auf Basis erfasster erster Eingangsdaten
  • - des ersten Maschinenprozesses, der den ersten Zustand aufweist, und/oder
  • - eines weiteren Maschinenprozesses, der den ersten Zustand aufweist, insbesondere des zweiten oder eines hiervon verschiedenen, insbesondere typgleichen oder typunterschiedlichen, Maschinenprozesses, der den ersten Zustand aufweist,
  und/oder
• - auf Basis erfasster zweiter Eingangsdaten
  • - dieses Maschinenprozesses, der den zweiten Zustand aufweist, trainiert, insbesondere also
• - auf Basis erfasster erster Eingangsdaten des ersten Maschinenprozesses, insbesondere der ersten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten des ersten Maschinenprozesses, insbesondere der ersten Maschine, der bzw. die den zweiten Zustand aufweist; und/oder
• - auf Basis erfasster erster Eingangsdaten des zweiten Maschinenprozesses, insbesondere der zweiten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten des zweiten Maschinenprozesses, insbesondere der zweiten Maschine, der bzw. die den zweiten Zustand aufweist; und/oder
• - auf Basis erfasster erster Eingangsdaten eines weiteren, vom ersten und gegebenenfalls zweiten Maschinenprozess verschiedenen (dritten) Maschinenprozesses, insbesondere wenigstens einer dritten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten dieses weiteren Maschinenprozesses, insbesondere wenigstens einer dritten Maschine, der bzw. die den zweiten Zustand aufweist.

In one embodiment, this network or GAN

• - based on recorded first input data
  • the first machine process having the first state, and / or
  • a further machine process having the first state, in particular the second or a different, in particular type-identical or type-different, machine process, which has the first state,
  and or
• - based on recorded second input data
  • - This machine process, which has the second state trained, in particular so
• - Based on detected first input data of the first machine process, in particular the first machine having the first state, and / or detected second input data of the first machine process, in particular the first machine having the second state; and or
• on the basis of detected first input data of the second machine process, in particular of the second machine, which has the first state, and / or detected second input data of the second machine process, in particular of the second machine, which has the second state; and or
• on the basis of detected first input data of a further, different from the first and possibly second machine process (third) machine process, in particular at least one third machine having the first state, and / or detected second input data of this further machine process, in particular at least one third Machine having the second state.

• - auf Basis erfasster erster Eingangsdaten
  • - des ersten Maschinenprozesses, der den ersten Zustand aufweist, und/oder
  • - eines weiteren Maschinenprozesses, der den ersten Zustand aufweist, insbesondere des zweiten oder eines hiervon verschiedenen, insbesondere typgleichen oder typunterschiedlichen, Maschinenprozesses, der den ersten Zustand aufweist,
  und/oder
• - auf Basis erfasster zweiter Eingangsdaten
  • - dieses Maschinenprozesses, der den zweiten Zustand aufweist,
  trainiert, insbesondere also
• - auf Basis erfasster erster Eingangsdaten des ersten Maschinenprozesses, insbesondere der ersten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten des ersten Maschinenprozesses, insbesondere der ersten Maschine, der bzw. die den zweiten Zustand aufweist; und/oder
• - auf Basis erfasster erster Eingangsdaten des zweiten Maschinenprozesses, insbesondere der zweiten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten des zweiten Maschinenprozesses, insbesondere der zweiten Maschine, der bzw. die den zweiten Zustand aufweist; und/oder
• - auf Basis erfasster erster Eingangsdaten eines weiteren, vom ersten und gegebenenfalls zweiten Maschinenprozess verschiedenen (dritten) Maschinenprozesses, insbesondere wenigstens einer dritten Maschine, der bzw. die den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten dieses weiteren Maschinenprozesses, insbesondere wenigstens einer dritten Maschine, der bzw. die den zweiten Zustand aufweist.

Additionally or alternatively, in one embodiment, the classification method

• - based on recorded first input data
  • the first machine process having the first state, and / or
  • a further machine process having the first state, in particular the second or a different, in particular type-identical or type-different, machine process, which has the first state,
  and or
• - based on recorded second input data
  • this machine process, which has the second state,
  trained, especially so
• - Based on detected first input data of the first machine process, in particular the first machine having the first state, and / or detected second input data of the first machine process, in particular the first machine having the second state; and or
• on the basis of detected first input data of the second machine process, in particular of the second machine, which has the first state, and / or detected second input data of the second machine process, in particular of the second machine, which has the second state; and or
• on the basis of detected first input data of a further, different from the first and possibly second machine process (third) machine process, in particular at least one third machine having the first state, and / or detected second input data of this further machine process, in particular at least one third Machine having the second state.

Thus, in one embodiment, the network having generating opposing subnetworks or GAN and the classification method with input data of the same machine process are trained.

Additionally or alternatively, in one embodiment (diagnostic) input data of that machine process is classified using the classification method with whose data the classification method and / or the network or GAN is or has been trained.

Additionally or alternatively, in one embodiment, fictitious input data are generated on the basis of detected first input data of that machine process on the basis of which (from its acquired first input data) the network is or has been trained with generating opposing subnetworks or GAN.

As a result, the network with generating opposing subnetworks or GAN or the classification method in one embodiment can be optimally adapted to the specific machine process.

In another embodiment, on the other hand, the network with generating opposing subnetworks or GAN and the classification method with input data of different machine processes are trained.

Additionally or alternatively, in one embodiment, input (diagnostic) input data of a machine process is classified using a classification method, wherein the classification method and / or the network or GAN is trained with input data of another machine process.

Additionally or alternatively, in one embodiment fictitious input data are generated on the basis of detected first input data, wherein the network is trained with generating opposing subnetworks or GAN with detected first input data of another machine process.

As a result, diagnostic input data of machine processes can also be classified in one embodiment, difficult or impossible to obtain for the first input data for training purposes, and / or poorly qualified (qualitatively and / or quantitatively), for example, noisy and / or only in small numbers.

Additionally or alternatively, in one embodiment, the (generated) fictitious input data additionally added (detected) second input data or added.

In this way, in one embodiment, the database for the second state can be (further) improved and / or enhanced, thereby improving the training of the classification method, in particular the training based on these additional (captured) second input data classification method diagnostic input data faster, more reliable and / or to classify more precisely.

Additionally or alternatively, in one embodiment, the network is trained with generating opposing subnetworks or GAN based on acquired second input data.

In this way, the training of the network or GAN can be improved, in particular the network or GAN trained on the basis of this additional (recorded) second input data can generate fictitious second input data faster, more reliably and / or more precisely.

In one embodiment, the number of first input data or examples with which the classification method is trained differs by at most 50%, in particular at most 35%, in an embodiment at most 25%, in particular at most 10%, of the number of notional input data or Sum of the fictitious and the second input data, with which the classification method is trained, from.

As a result, in one embodiment the classification method can be trained particularly advantageously, in particular the classification method trained on the basis of this additionally (generated fictitious input data) can classify diagnostic input data faster, more reliably and / or more precisely.

In one embodiment, the first machine process, the second machine process and / or the further or third machine process, in particular one or more of its components, in particular machines, specifically transferred from the first to the second state to capture second input data, which in a Execution for training the classification method and / or the network with generating opposing subnetworks or GAN be used, in particular by specifically using at least one faulty component, in particular installed or used, is.

In this way, in one embodiment, the database for the second state can be (further) improved and / or increased and thereby the training of the classification process or network or GAN can be improved, in particular the classification method can classify diagnostic input data faster, more reliably and / or more precisely.

An inventive classification of diagnostic input data of a machine process can be used with particular advantage for the diagnosis of the machine process, without the present invention being limited thereto. Rather, the classification in one embodiment can also be used for controlling the machine process, in particular one or more of its components, in particular machines or else for other purposes.

• - Erfassen von Diagnose-Eingangsdaten des Maschinenprozesses;
• - Klassifizieren dieser Diagnose-Eingangsdaten mithilfe des trainierten Klassifikationsverfahrens in einer hier beschriebenen Weise; und
• - Bewerten des Maschinenprozesses auf Basis dieser Klassifizierung.

According to an embodiment of the present invention, a method for diagnosing a machine process comprises the steps of:

• - Acquisition of diagnostic input data of the machine process;
• Classifying these diagnostic input data using the trained classification method in a manner described herein; and
• - Evaluate the machine process based on this classification.

In one embodiment, a diagnosis or an evaluation of a machine process can include, in particular, monitoring and / or forecasting (a state and / or work result) of the machine process, in particular one or more of its components, in particular machines. In one embodiment, on the basis of the diagnosis or of the evaluated machine process, a, in particular prospective, maintenance (s) or maintenance (sstrategy) of the machine process, in particular of one or more of its components, in particular machines, is planned ("predictive maintenance"), which may represent a particularly advantageous application of the present invention. Additionally or alternatively In one embodiment, work results of the machine process can be handled differently on the basis of the diagnosis or of the evaluated machine process, for example rejected rejects or fed to a post-processing (quality assurance).

• - Mittel zum Generieren von fiktiven Eingangsdaten eines ersten Maschinenprozesses, insbesondere wenigstens einer ersten Maschine, in einer Ausführung wenigstens eines ersten Roboters, der bzw. die einen zweiten Zustand aufweist, auf Basis erfasster erster Eingangsdaten dieses Maschinenprozesses bzw. dieser Maschine, der bzw. die einen ersten Zustand aufweist, mittels maschinellem Lernen;
• - Mittel zum Trainieren eines Klassifikationsverfahrens auf Basis der (generierten) fiktiven und (erfassten) ersten Eingangsdaten; und
• - Mittel zum Klassifizieren von Diagnose-Eingangsdaten des ersten und/oder wenigstens eines, insbesondere typgleichen oder typunterschiedlichen, zweiten Maschinenprozesses, insbesondere der ersten und/oder wenigstens einer, insbesondere typgleichen oder typunterschiedlichen, zweiten Maschine, in einer Ausführung des ersten und/oder wenigstens eines, insbesondere typgleichen oder typunterschiedlichen, zweiten Roboters, mithilfe des trainierten Klassifikationsverfahrens.

According to one embodiment of the present invention, a system, in particular hardware and / or software, in particular program technology, is set up to carry out a method described here and / or has:

• Means for generating fictitious input data of a first machine process, in particular of at least one first machine, in an embodiment of at least one first robot, which has a second state, based on detected first input data of this machine process or machine, the or the having a first state by machine learning;
• - means for training a classification method based on the (generated) fictitious and (detected) first input data; and
• Means for classifying diagnostic input data of the first and / or at least one, in particular type-identical or type-different, second machine process, in particular of the first and / or at least one, in particular type-identical or type-different second machine, in one embodiment of the first and / or at least one, in particular type-identical or type-different, second robot, using the trained classification method.

• Mittel zum Generieren der fiktiven Eingangsdaten mithilfe eines Netzwerks mit erzeugenden gegnerischen Subnetzwerken; und/oder
• Mittel zum Trainieren des Netzwerks und/oder des Klassifikationsverfahrens auf Basis erfasster erster Eingangsdaten des ersten und/oder eines weiteren, insbesondere des zweiten oder eines hiervon verschiedenen, Maschinenprozesses, der den ersten Zustand aufweist, und/oder erfasster zweiter Eingangsdaten dieses Maschinenprozesses, der den zweiten Zustand aufweist; und/oder
• Mittel zum gezielten Überführen des ersten, zweiten und/oder weiteren Maschinenprozesses von dem ersten in den zweiten Zustand, um zweite Eingangsdaten zu erfassen; und/oder
• Mittel zum Erfassen von Diagnose-Eingangsdaten eines Maschinenprozesses, Klassifizieren dieser Diagnose-Eingangsdaten mithilfe des trainierten Klassifikationsverfahrens in hier beschriebener Weise und Bewerten dieses Maschinenprozesses auf Basis dieser Klassifizierung.

In one embodiment, the system or its agent has:

• Means for generating the notional input data using a network of generating opposing subnetworks; and or
• Means for training the network and / or the classification method based on detected first input data of the first and / or another, in particular the second or a different, machine process having the first state, and / or detected second input data of this machine process, the second state; and or
• Means for selectively transferring the first, second, and / or further engine processes from the first to the second state to acquire second input data; and or
• Means for acquiring diagnostic input data of a machine process, classifying said diagnostic input data using the trained classification method as described herein, and evaluating that machine process based on that classification.

An agent in the sense of the present invention may be designed as a hardware and / or software, in particular a data or signal-connected, preferably digital, processing, in particular microprocessor unit (CPU), graphics card (GPU), preferably with a memory and / or bus system ) or the like, and / or one or more programs or program modules. The processing unit may be configured to execute instructions implemented as a program stored in a memory system, to acquire input signals from a data bus and / or to output output signals to a data bus. A storage system may comprise one or more, in particular different, storage media, in particular optical, magnetic, solid state and / or other non-volatile media. The program may be arranged to embody or perform the methods described herein so that the processing unit may perform the steps of such methods and, in particular, diagnose the machine process. In one embodiment, a computer program product may include, in particular, a non-volatile storage medium for storing a program or a program stored thereon, wherein execution of this program causes a system or a controller, in particular a computer, to do so method described herein or one or more of its steps.

In one embodiment, one or more, in particular all, steps of the method are completely or partially automated, in particular by the system or its (e) means. In one embodiment, the system comprises the machine (s), in particular robots, of the machine process.

• 1: ein System zur Klassifikation von Eingangsdaten eines Maschinenprozesses nach einer Ausführung der vorliegenden Erfindung; und
• 2: ein Verfahren zur Klassifikation von Eingangsdaten des Maschinenprozesses nach einer Ausführung der vorliegenden Erfindung.

Further advantages and features emerge from the subclaims and the exemplary embodiments. This shows, partially schematized:

• 1 a system for classifying input data of a machine process according to an embodiment of the present invention; and
• 2 A method of classifying input data of the machine process according to an embodiment of the present invention.

1 shows a system for the classification of operating parameters of a machine process of a robot 1 with a robot controller 2 . 2 a method for the classification of operating parameters of the machine process or robot 1 according to an embodiment of the present invention.

In a first step S10 become first operating parameters X ₁ of the robot 1 , which during and during this has a first, error-free or "good" state ("IO"), as well as second operating parameters X ₂ of the robot 1 , which has a second, bad state ("NIO"), and thereby a network N trains, the generating enemy subnetworks N ₁ . N ₂ having.

In a second step S20 generates the network N fictitious operating parameters F of the robot 1 if it has one or the second, bad state ("NIO"). In other words, the network emulates N Operating parameters, as in a faulty or bad state of the robot 1 are detected, or pretend such.

With these fictitious operating parameters F and the first operating parameters X ₁ will be in step S30 a classification procedure K trained.

In this case, the detected second operating parameters X ₂ the fictitious operating parameters F added and the classification method K thus also with these detected second operating parameters X ₂ trained.

To capture the second operating parameters X ₂ can the robot 1 be temporarily transferred to the second, incorrect or bad state, for example, by selectively a defective transmission is installed or the like.

In one step S40 become diagnostic operating parameters X _D of the robot 1 detected.

In one step S50 classify that in step S30 trained classification methods K these diagnostic operating parameters X _D ,

Based on this classification will be in one step S60 the robot 1 monitors and plans a maintenance (maintenance) strategy.

Although exemplary embodiments have been explained in the foregoing description, it should be understood that a variety of modifications are possible.

Thus, in the embodiment, the network N with operating parameters X ₁ and X ₂ of the robot 1 then trains and generates fictive operating parameters F this robot. In a modification, the network can N but also with operating parameters of another robot of the same or another type (not shown) and then fictitious operating parameters of the robot 1 to generate.

Similarly, in the embodiment, the classification method K with fictitious operating parameters F as well as operating parameters X ₁ and X ₂ of the robot 1 then trains and classifies diagnostic operating parameters X _D this robot. In a modification, the network can N but also with fictitious and first and second operating parameters of another robot of the same or another type (not shown) are trained and then diagnostic operating parameters of the robot 1 classify.

It should also be noted that the exemplary embodiments are merely examples that are not intended to limit the scope, applications and construction in any way. Rather, the expert is given by the preceding description, a guide for the implementation of at least one exemplary embodiment, with various changes, in particular with regard to the function and arrangement of the components described, can be made without departing from the scope, as it turns out according to the claims and these equivalent combinations of features.

LIST OF REFERENCE NUMBERS

1

robot

2

control

F

fictitious input data

K

classification methods

N _(1/2)

(Sub) network

X ₁

first input data

X ₂

second input data

X _D

Diagnostic input data

Claims (8)

Method for classifying input data (X _D ) of a machine process, in particular of a machine (1), with the steps: - generating (S20) fictitious input data (F) of a first machine process having a faulty second state based on detected first input data (X ₁ ) of this machine process having a first condition free from errors, by means of machine learning; - training (S30) a classification method (K) on the basis of the fictitious and first input data; and - Classifying (S50) diagnostic input data (X _D ) of the first and / or at least one second machine process by means of the trained classification method, the notional input data being generated by means of a network (N) having generating opposing subnetworks (N ₁ , N ₂ ). Method according to the preceding claim, characterized in that the network (N) and / or the classification method (K) based on detected first input data of the first and / or another, in particular the second or a different, machine process, the first state , and / or detected second input data of this engine process having the second state is trained (S10, S30). Method according to one of the two preceding claims, characterized in that the number of the first input data, with which the classification method is trained by at most 50% of the number of notional input data or the sum of the notional and the second input data, with which the classification method trains will, deviates. Method according to one of the two preceding claims, characterized in that the first, second and / or further machine process is selectively transferred from the first to the second state to detect second input data. Method for diagnosing a machine process, comprising the steps of: - detecting (S40) diagnostic input data of the engine process; Classifying (S50) said diagnostic input data using the trained classification method of any one of the preceding claims; and - Evaluate (S60) the machine process based on this classification. Method according to one of the preceding claims, characterized in that the input data operating parameters of at least one machine, in particular at least one robot. System for classifying input data of a machine process, in particular a machine (1), which is set up to carry out a method according to one of the preceding claims. A computer program product having a program code stored on a computer-readable medium for performing a method according to any one of the preceding claims.

Images