Method, (measuring) device and components thereof for detecting events in a material processing or manufacturing process using event patterns

Abstract

The invention relates to a measuring device comprising a sensor for high-frequency detection of measured variables, such as sound, structure-borne sound, current, voltage, optical or magnetic measured values ​​and the like, the measuring device having an operator network with at least one operator. The invention further relates to an operator network having a signaling operator, a machine control operator and / or a difference operator, and the invention relates to said operator. The invention further relates to a method or a device for identifying events in a material processing and / or manufacturing process based on a multidimensional data stream with time-resolved frequency information and energy information obtained during a process and based on the use of interference patterns and / or event patterns and / or energy data, wherein the data are fed to a pattern operator or to a plurality of pattern operators arranged in parallel or in succession.

Description

Technical Field

[0001] The invention relates to a method, a (measuring) device and components thereof for identifying events in a material processing process or a material manufacturing process using event patterns based on a multidimensional data stream with time-resolved frequency information and energy information obtained during the process.

[0002] The invention is applicable to all processes in which information can be obtained at high frequency by means of, in particular, acoustic and / or magnetic sensors during and / or after material processing or production, as described, for example, in WO 2010 / 051954 and WO 2017 / 071812, for example in machining, in injection molding, in forming, in induction hardening, in welding or the like. For descriptive purposes, the invention is described below using the example of bend straightening. However, the described embodiments can also be transferred to other applications. Background Art

[0003] During bend straightening on a straightening press, the punch acts repeatedly on the component to be straightened for plastic deformation and at different locations until the desired runout is achieved. For example, a typical bend straightening process for a transmission shaft made of hardened steel used in the automotive industry can last for about one or two minutes, including a load phase that is typically in the order of magnitude of less than 1 second and a subsequent unloading phase of substantially similar length. The deformation energy acting on the component can cause damage to the component that is not easily detectable optically, in particular microcracks, and the bend straightening device can also be damaged or worn. In order to detect such damage during a material processing process or a material manufacturing process, it is known from WO 2010 / 051954 to continuously detect sound signals during the bend straightening process with the help of a sound sensor and to feed them to a multidimensional evaluation with pattern recognition in order to detect typical events, such as cracks or tool breakage.

[0004] Figure 1 The data stream detected here is explained as a function of time t, frequency f and intensity I, as explained in WO 2010 / 051954. The landscape (Lanschaft), which is explained here in three dimensions by way of example, has patterns that can be assigned to different events.

[0005] Thus, in loading phase 1 a picture deviating from the picture in loading phase 2 or in unloading phase 3 can be seen.

[0006] If the profile is compared with a known pattern, for example, for cracks, it can be determined whether the component is defect-free or defective and therefore rejectable. If the profile 4 in the region of load phase 1 corresponds, for example, to a pattern for cracks, the component is classified as defective. If, however, the profile 4 corresponds to a pattern for operating noises, as in the region of the preceding load phase 2, the component is defect-free.

[0007] What is difficult here is an exact distinction. Thus, although the degree of agreement can be determined by simple pattern recognition, for example by comparing the recorded images with the known comparison envelope of the comparison pattern, it has been shown in practice that the agreement is often significantly below 100% and that multiple patterns for different events with the same degree of agreement in the same order of magnitude may be identified, for example, with an 80% probability for one pattern and a 65% probability for another pattern. Depending on the quality requirements for the component, this can result in a non-defective component being incorrectly classified as defective or a defective component being passed through as non-defective.

[0008] Furthermore, the adaptation to the respective process is complex. Thus, a database for the process and the respective process environment must be provided and / or adapted. A specific adaptation of the comparison logic also usually requires reprogramming. Thus, the modes for tolerances and for interference or operating noise in bend straightening differ significantly from those in injection molding; the optimized comparison logic is also always different for different processes. Summary of the invention

[0009] Based on this, the object of the present invention is to provide a method, a device and its components for identifying events in a material processing process or a material manufacturing process based on a multidimensional data stream with time-resolved frequency information and energy information obtained during the process using event patterns, wherein the method, the device and its components are able to achieve improved differentiation with more flexible adaptability.

[0010] This object is achieved according to the features of the independent claims.

[0011] Advantageous embodiments are revealed in the dependent claims.

[0012] The present invention provides a measuring device, such as a measuring instrument, which has an operator network that can be easily adapted to a corresponding material processing process and / or manufacturing process, advantageously via a graphical user interface or the like, the operator network having at least one determination operator and preferably additional operators, in particular arranged in series and / or parallel, in particular pattern operators, and the present invention provides a method and device constructed thereon.

[0013] In some embodiments, the measuring device according to the invention enables a fully automatic decision, for example OK or not OK, during or after material or component processing or manufacture, and can be intuitively, simply and precisely adapted to the corresponding material, component, manufacturing and / or processing framework conditions, advantageously using predetermined and / or parameterizable operators and / or operator networks and optionally using predetermined standard parameters and / or standard patterns for specific material, component, manufacturing and / or processing standard framework conditions or situations. In particular, it can be provided that the pre-determination as a template is adapted in this way by means of characteristic curves / parameterizations, further operators and their associations, and / or redefinition of operators.

[0014] The measuring device according to the invention can include sensors for high-frequency detection of measured variables, such as sound, structure-borne sound, current, voltage, optical or magnetic measured values ​​and the like, or it can be coupled to the sensor and have an arithmetic unit which can be integrated into an arithmetic unit network.

[0015] The computing unit system according to the invention essentially forms a dedicated configuration environment or programming language, in particular for measuring instruments. Specific sensors or general computing units are provided here, which provide measured values ​​for the computing unit to be processed.

[0016] Sensors that support existing communication protocols, such as OPC-UA, can be integrated via the OPC-UA operator.

[0017] In this way, almost any number of different sensors can be included in the programmed evaluation.

[0018] Such programmable pattern recognition allows any signals to be identified, their specific characteristics to be calculated and removed from the data stream in real time, which signals may be machine noise and well-known operating emissions in the case of sound or structure-borne noise.

[0019] This results in a highly dynamic signal or noise filter which meets all criteria for a noise elimination function.

[0020] The invention thus enables a measuring instrument to precisely remove extremely "loud" elementary signals from a data stream and then to identify small features in the "remaining" data stream.

[0021] According to the invention, this can be solved in that a pattern operator searches the input sensor data stream for known noise, then removes the noise and generates a new data stream, which can be analyzed and processed by subsequent and / or parallel further pattern operators until the result is analyzed to be usable for decision (decision maturity, entscheidungsreif) and can be supplied to the decision operator.

[0022] Such a decision operator is designed according to the invention for determining, for example, whether a component is OK or not OK in a material processing process and / or a material manufacturing process (or the like), or whether a processing device and / or a manufacturing device is OK or not OK. The decision operator has an input for an interference pattern recognition operator, an input for an event pattern recognition operator (8) and / or an input for a free pattern recognition operator (9), and optionally also has multiple inputs for example for multiple event pattern recognition operators and / or multiple inputs for example for multiple free pattern recognition operators. Internal logic and internal parameters are suitably provided for parameterizing and programming the input data, as well as an output for an OK signal or a not OK signal.

[0023] The decision operator can include a multi-dimensional, in particular a two-dimensional, three-dimensional, four-dimensional or five-dimensional decision field, in particular a decision field having one or more associated regions.

[0024] The decision computing unit can have a plurality, in particular two, three or four decision fields, each of which has at least one temporally variable boundary line.

[0025] In addition to the decision computing unit and / or the pattern recognition computing unit, a further computing unit is expediently provided for communication with the outside world, in an industrial environment with a machine control unit or a higher-level data system.

[0026] The pattern recognition operator according to the invention can have an input for the data stream and an output for the data stream and / or a value which corresponds to the degree of agreement with a pattern from a plurality of patterns in the pattern database.

[0027] The pattern recognition operator expediently recognizes event patterns and / or interference patterns, in particular by means of parameterization, decision fields and / or pattern recognition.

[0028] The pattern recognition operator expediently has a two-dimensional, three-dimensional or multi-dimensional characteristic curve having one or more associated regions.

[0029] A plurality of operators form an operator network according to the invention, which comprises one, two or more pattern recognition operators (7, 8), optionally one or more free pattern operators (9) and at least one decision operator (6) and / or further operators. The operator network can have a notification operator, a machine control operator and / or a difference operator.

[0030] In this case, a plurality of operators, in particular mode operators, can be arranged in parallel or in series in the operator network.

[0031] Multiple data transformations that are suitably arranged allow for the step-by-step processing of the data, up to two-dimensional or multi-dimensional diagrams, several values ​​and yes / no decisions. Thus, by using computer programming and possibly using different external sensors, completely different measuring instruments can be created simply by implementing different evaluation algorithms, which are inserted into the computer network by drag and drop.

[0032] The parameterizations obtained once for the individual operators, which can be very complex, such as in the case of a pattern search or a decision search, are advantageously stored as template parameterizations and can be used again in other measurement contexts.

[0033] Complete operator networks can also be stored as template networks and made available as a basis for similar or expanded measuring systems.

[0034] The computing device network can be implemented on all compatible hardware systems (measuring instruments) and thus forms a general analysis layer and even a usable measuring instrument programming for graphical representation.

[0035] The computing units are connected by connections, also called connections, which transmit the correspondingly processed data between the individual computing units.

[0036] These connections can also be parameterized as buffer memories with the Pipeline function. The connections thus ensure asynchronous processing of data, which is necessary in particular when switching between operators or data streams that operate at different speeds.

[0037] The connection thus also makes it possible to connect a plurality of distributed computing networks, which are connected on different hardware via the Internet to an industrial bus system or in some other way, to form a complex evaluation network.

[0038] The mobile device supporting the computing device according to the invention can be part of an integrated computing device network or can be a sensor or a display module.

[0039] The invention therefore provides a mobile, in particular electrically autonomous device, in particular in the form of a sensor or display module, which is designed as a computing unit with an input data stream and / or an output data stream for embedding in a computing unit network or a measuring device. Communication takes place in a known manner, either wired or wirelessly, via standard protocols such as TCP / IP or the like.

[0040] The present invention therefore provides methods and devices, such as measuring instruments, for identifying events in a material processing process and / or a material manufacturing process based on a multidimensional data stream with time-resolved frequency information and energy information obtained during a process (for example, from a structure-borne sound sensor or other mentioned sensors) using interference patterns and / or event patterns and / or energy data, wherein the data pass through a pattern operator or multiple pattern operators arranged in parallel or in series and are supplied to a decision operator or multiple decision operators arranged in parallel or in series.

[0041] The invention further provides a graphical user interface for constructing and / or editing an inventive computing unit network and / or a measuring device comprising a graphical user interface, wherein graphical localization and associatability of predetermined computing units are possible.

[0042] In the device according to the invention for carrying out the method, a network of operators and / or a graphical user interface for constructing and / or editing the network of operators is provided.

[0043] Further features and embodiments emerge from the subsequent description of exemplary embodiments of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Aspects of the invention are described below with reference to the embodiments illustrated in the drawings. DETAILED DESCRIPTION

[0045] The operator network 5 according to the present invention is configured in an exemplary manner. Figure 2 The operator network comprises a decision operator 6, which is connected here at the input to a disturbance mode operator 7, an event mode operator 8 and / or a free mode operator 9 and at the output to a signal operator 10, a counter operator 11 and / or a storage operator 12.

[0046] Each operator here is a program module with a predetermined input interface and / or output interface and an internal logic, which is used to filter data, detect data, process data, make decisions, store data, interact with computers, databases, users and / or the like and / or activate devices (such as signals, motors, cutting devices and / or the like).

[0047] The interference pattern operator 7 shown here can be as shown in Figure 3The data stream 13, for example as a raw data stream of a sensor, in particular a sound signal sensor, or as a data stream processed, for example frequency filtered and / or smoothed, in particular by other operators, arrives at a position determination operator 14, which determines the time position of the data within the process, taking into account the process time provided by a time operator 15, and supplies the time position to a pattern comparison operator 16. An energy determination operator 17 is likewise connected to the operators 13, 15, determines the energy in the data stream, for example in the form of sound intensity, and relates it to the process time in order to supply the result to the pattern comparison operator 16. With the aid of these two input data streams and optionally, as shown, a data stream from a data stream operator 13, a time component from a time operator 15 and / or pattern data from a pattern operator 18, a pattern comparison operator 16 determines a degree of agreement or a plurality of degrees for a plurality of patterns for the data provided by the data stream operator 13 and the pattern provided by the pattern operator 18, which is advantageously connected to a pattern database or can call up patterns for events in another way. The degree can be, for example, a numerical value between 0% and 100% or between 0 and 255 or the like. The degree of agreement is provided at a result output 19.

[0048] The pattern calculator 18 accesses interference patterns, such as those occurring during the process, such as impacts of the punch on the component to be straightened, humming noises due to vibrations of the straightening device, environmental influences (e.g. knocks by the operator), grinding noises of the tool on the workpiece, etc. At the result output 19, for example, a probability of 67% that an interference pattern has been identified can be output.

[0049] The event pattern operator 8 can be constructed in essentially the same way as the interference pattern operator 7. Thus, the logic of the event pattern operator can also include a data flow operator, a time operator, a position determination operator, an energy determination operator, a pattern comparison operator and / or a pattern operator, which can be connected similarly or identically and configured internally with weighting factors and the like, wherein the pattern operator accesses event patterns from a pattern library, which event patterns describe interferences, such as patterns for stress cracks or tool fractures during cutting processes and the like. The event pattern operator 8 outputs a value representing that such an event has occurred, for example 82%, at its result output 19, see Figure 2 .

[0050] The free mode operator 9 can be as shown in Figure 4 As shown in Figure 3In addition to the data flow operator 13, the time operator 15 and the position determination operator 14, there is also an energy determination operator 20, which calculates and provides energy, such as sound signal strength and / or duration, at its result output 19. Instead of pattern recognition, an energy analysis is performed here.

[0051] The process and / or the result can be advantageously displayed on a screen or the like by means of a visualization unit 21 .

[0052] According to the invention, the computing units can be arranged and configured in a graphical user interface. Thus, the computing units can be pulled out of the computing unit list and arranged in a working page by dragging and dropping with a mouse or the like. Figure 2 , 3 4 in the form of lines between the input and output can also be configured graphically. By right-clicking on the operator or the like, internal parameters or logic of the operator can be selected and / or changed. Thus, for example, a formula for calculating an energy value can be determined in the energy determination operator 20, an algorithm can be selected in the pattern comparison operator 16, with the aid of which the degree of agreement between the detected data and a certain pattern is determined (for example, by accumulation of differences or the like), etc.

[0053] The graphical user interface enables the user to quickly and flexibly implement, adapt and maintain a computing device network specifically coordinated to the respective process, without being dependent on the manufacturer or programmer.

[0054] It is of particular importance for the configuration of the decision computing unit 6 that it is able to distinguish between defective (OK) and defective (not OK, NOK) components / workpieces in a reliable manner.

[0055] exist Figure 2 In the example shown, the OK input 22 , the NOK input 23 and the NO input 24 of the decision operator 6 are connected to the result outputs 19 of the operators 7 , 8 and 9 .

[0056] At the OK input 22, the interference pattern operator 7 supplies a (probability) value representing that an insignificant interference signal is currently detected, i.e. the workpiece is "OK", together with further information. At the NOK input 23, the event pattern operator 8 supplies a (probability) value representing that a significant event (e.g. a stress crack) is currently detected, i.e. the workpiece is not OK ("NOK"), together with further information. At the NO input 24, further information is supplied without pattern comparison ("NO" pattern recognition). The further information can here include energy values, such as sound intensity and / or duration, such as the duration of the energy from exceeding a threshold value until it falls below a threshold value, or the like.

[0057] According to the invention, the decision operator 6 is configured such that it contains a plurality of multidimensional decision fields in series and sends a trigger signal to the signal operator 10 based on the result of the series multidimensional decision, so as to characterize the workpiece as defective (or optionally non-defective), so that defective workpieces can be extracted. Likewise, the results and / or data can be stored by forwarding to the storage operator 12, and the rejects can be detected per time unit or overall by the counting operator 11 and displayed on the visualization operator 21, for example in the form of a scale, in order to graphically show the trend.

[0058] The first judgment field of multidimensional Figure 5 The first judgment field records the OK value from the interference operator 7 and the NOK value from the event pattern operator 8. In the example mentioned above, Figure 1 In load phase 1, 67% OK values ​​and 82% NOK values ​​were determined. This corresponds to data point 25, which is above the boundary line 26 that can be configured via the graphical user interface. Another data point 27 is below the boundary line 26. Data points 25, 26 do not involve a clear assignment, i.e. the event can be an insignificant interference signal or a significant event pattern signal with a high probability, while other data points 28, 29 are clear because they are either 0% NOK or 0% OK.

[0059] Boundary line 26 is used here as a first level of differentiation. Ambiguous data points are fed to the next level according to Figure 6 The multi-dimensional NOK judgment field, according to Figure 7 The multi-dimensional OK judgment field and / or according to Figure 8 Multi-dimensional NO judgment field.

[0060] Thus, the data point 25 is accordingly recorded with its further data, here by way of example energy and duration, and is Figure 5 Located above the NOK border line 30, Figure 6 is below the OK boundary line 31 and Figure 7 In other embodiments, the two-dimensional or more dimensional decision field may include parameters such as energy, duration, position in time, counter, time scale, amplitude scale and the like and / or boundary surfaces may be used instead of boundary lines.

[0061] The decision operator 6 is expediently configured in such a way that it evaluates the data point 25 as NOK, since the data point 25 is located in the NOK region 32 with respect to any two of the three operators OK, NOK and NO.

[0062] The boundary lines 26 , 30 , 31 , 33 , 34 can be specifically adapted to the respective process.

[0063] exist Figure 5 , 6 Another dimension in , 7, 8 can be time. It is thus expedient to adjust the boundary lines 26, 30, 31, 33, 34 in the process as a function of time. For example, at the beginning of each load phase 1, 3, the force applied to the workpiece during bend straightening is small, and therefore the probability of fracture is small, while the force or the probability of fracture is maximum at the end of the respective load phase 1, 3. It can thus be expedient to set a boundary line 26' that is larger than the boundary line 26 at the beginning of the load phase, and to set a boundary line 26" at the end of the load phase, in which the NOK event is started earlier due to the high forces, see Figure 5 . Thus, data point 27 can be OK when it is detected at the beginning of the load phase, while it is above boundary line 26" at the end and is NOK. The time curve between boundary line 26' at t=beginning of the load phase (0 ms) via boundary line 26 to boundary line 26" at t=end of the load phase (e.g. 1000 ms) can be adjusted arbitrarily, for example linearly.

[0064] The above-mentioned operators can be arbitrarily connected in series and / or in parallel.

[0065] Thus, it is expedient to provide a further decision operator 6', see Fig. 9 , the further decision computing unit monitors the maintenance state of the bending device. Thus, the pattern for the event can be evaluated here, for example Figure 7 Data points with high energy and long duration can infer a problem with the machine and may prescribe stopping the machine.

[0066] A further computing unit may be provided, for example a notification computing unit, which automatically sends an e-mail or another message with the relevant information to a maintenance service and / or a process agent.

[0067] In another embodiment, a difference operator can be provided. The difference operator is configured in such a way that it detects known patterns, such as more or less constant background noise or recurring interference noise or other interference data, adapts the associated pattern as accurately as possible to the order of magnitude of the data stream and subtracts it from the data stream if it is sufficiently consistent. The cleaned data stream then remains as a difference and can be fed to a further operator. A possible arrangement is in Fig.10 In this arrangement, the difference operator 35 is connected downstream of the interference pattern operator 7' and extracts the interference pattern from the data stream 13 based on the interference pattern recognition and provides the cleaned data stream 36 for further processing, for example as a signal for the operators 7, 8, 9. Figure 2 , wherein the subsequent operator 7 - if present - can be adjusted differently, or the interference pattern operator 7 ′ only detects certain interference patterns, such as ripples, scattering, etc.

[0068] The computing machine model according to the invention is explained below with reference to further examples. In particular, the computing machine model can be implemented on distributed, heterogeneous and asynchronously operating hardware and systems of information technology.

[0069] In the measuring device according to the invention, the evaluation is carried out via a network of computing units.

[0070] At the beginning there is an operator that enables access to input data.

[0071] The measurement data stream from the external source can sample the sensor data 4 through the channel with up to 100M samples (Msamples) and 16 bits or 4M and 24 bits and calculate the FFT at 25000 spectra / second per channel in real time and provide both as input data streams.

[0072] 32-bit IO for machine control as well as Profibus (fieldbus) and Profinet for machine communication can also be provided as input and output for the calculation unit.

[0073] The analysis process is constructed in the graphical operator editor according to the invention, see for example Fig.11 .

[0074] In the top layer, the processor (Clock Slot) executes the analysis program (custom operator).

[0075] Custom operators form containers for subroutines in which complex networks from other subroutines and prefabricated operators can be implemented, see Fig.12, which exemplarily shows a network for pattern analysis and decision making.

[0076] On the next layer we can see a network combining data flow, filtering components and pattern comparison operations, see Fig.13 .

[0077] Each operator presents a more or less complex GUI component in the editor, which defines the connection to the data and parameterizes the way the operator works.

[0078] Data in the form of some primitive types, booleans, numbers, int64, etc., and more complex data, for example in the form of JSON objects, are exchanged over the connection.

[0079] For parameterization, JSON objects are also communicated to the GUI.

[0080] Integration of new operators can be done via plugins.

[0081] That is, a new data source must first bring in the operator plugin.

[0082] During real-time measurement, the data are processed by an operator and, as shown in this example, the best matching pattern is found, scaled and its counterpart (Pendant) in the data stream is analyzed.

[0083] At the same time, an evaluation is carried out according to the requirements of the process and a decision is made, for example, with regard to the workpiece quality or the performance of the production process.

[0084] exist Fig.14 In the case of a micro crack in a drive shaft, for example, a 50 ms long section of structure-borne sound data from an industrial straightening machine for drive shafts is shown. The image is generated based on 25,000 spectra / second according to the FFT of the microphone data, which in this case were evaluated up to approximately 800 kHz.

[0085] All data can be stored in the measuring instrument in RAW format and with different compressions and analyzed both in real time and after the process (for learning processes).

[0086] Examples of sensors according to the invention may be structure-borne sound microphones and μ-magnetic sensors as well as broadband analog amplifiers (>100 MHz) for data acquisition in industrial processes.

[0087] The design of computing machines can be extended to distributed, heterogeneous and asynchronous hardware and systems of information technology. Parts of large computing machine networks can run on distributed systems.

[0088] The sensor according to the invention can collect and preprocess data, ie is designed as an intelligent sensor.

[0089] Real-time processing systems can perform spectrum analysis and recognize patterns. PCs make complex decisions. The hardware and software used can be very different. Any device, any service can be integrated into the overall execution as long as the interface of the computing system is supported.

[0090] Manufacturers of devices and smart sensors provide, in addition to the actual devices, modules for the integration of computing units, which can be parameterized via a GUI and provide inputs and outputs for other participants in the network.

[0091] In this case, the data stream can be both analog and digital and can be transmitted on different carriers and can also support different protocols, the corresponding operators of which provide access to the data in a compatible form.

[0092] In all embodiments, a learning phase can be provided, in which the sensor can transmit a large amount of data, and the learning is slow. Based on the learning data, an analysis network can be constructed and parameterized. The calculation part is then carried out in the sensor itself, the data is already pre-processed and evaluated, and in the best case the result is already transmitted.

[0093] The system with wireless energy supply and wireless communication can be designed to be completely closed and therefore process-safe, and can be used without maintenance.

[0094] For network integration, an OS (e.g. Linux) with a TCP / IP connection can be run on each sensor. Alternatively, a sensor hub can be set up that contains the OS system and places the sensors in the network. Forwarding can be wireless or wired.

[0095] In order to supply power to wireless components such as sensors, according to the invention, preferably in addition to battery power supply, methods of energy harvesting and wireless energy transmission are used.

[0096] In an advantageous embodiment, if the data are provided by the sensor system primarily via TCP / IP, an intelligent switch based on an FPGA, for example, can form the collection point and the further processing stage. The intelligent switch performs calculations on the connected data streams in parallel and then provides its results again either via TCP / IP or via PCIe, USB, etc.

[0097] Reference numerals list

[0098] 1 Load phase

[0099] 2 Load phase

[0100] 3 Uninstallation phase

[0101] 4. Image

[0102] 5. Operator Network

[0103] 6.6' Determination operator

[0104] 7 Interference Pattern Operator

[0105] 8 Event Pattern Operator

[0106] 9 Free Mode Operator

[0107] 10 Signal operator

[0108] 11 Counting unit

[0109] 12 Storage Arithmetic Unit

[0110] 13 Data Flow Operator

[0111] 14 Position determination operator

[0112] 15 Time calculator

[0113] 16-mode comparison operator

[0114] 17 Energy determination operator

[0115] 18 Mode operator

[0116] 19 Result output

[0117] 20 Energy determination operator

[0118] 21 Visualization Component

[0119] 22 OK input

[0120] 23 NOK input terminal

[0121] 24 NO input

[0122] 25 data points

[0123] 26, 26', 26" Boundary Line

[0124] 27 data points

[0125] 28 data points

[0126] 29 data points

[0127] 30 NOK border line

[0128] 31 OK border line

[0129] 32 NOK Area

[0130] 33 NO Boundary Line

[0131] 34 NO Boundary Line

[0132] 35 Difference operator

[0133] 36 Data Stream

Claims

1. A measuring device comprising a sensor for high-frequency detection of a measurement variable, characterized in that The measuring device comprises an operator network having: one, two or more pattern recognition operators having an input for the data stream and an output for the data stream and / or a value corresponding to a degree of conformity with one of a plurality of patterns in a pattern database; as well as At least one decision operator (6), which is used to determine whether a component is OK or not OK in a material processing process and / or a material manufacturing process, or to determine whether a processing device and / or a manufacturing device is OK or not OK, the decision operator having an input for an interference pattern recognition operator (7), an input for an event pattern recognition operator (8) and / or an input for a free pattern operator (9), the decision operator having internal logic and internal parameters for parameterizable and / or programmable association of input data, and the decision operator having an output for an OK signal or a not OK signal, wherein the decision operator includes a multi-dimensional decision field, the pattern recognition operator having a two-dimensional, three-dimensional or more dimensional characteristic curve, and the characteristic curve having one or more assigned areas.

2. The measuring device according to claim 1, characterized in that The sensor is designed for high-frequency detection of sound, structure-borne sound, current, voltage, optical measured values ​​or magnetic measured values.

3. The measuring device according to claim 1, characterized in that The decision operator includes a two-dimensional or three-dimensional decision field.

4. The measuring device according to claim 3, characterized in that The decision operator comprises three decision fields, each of which has at least one temporally variable boundary line (26, 26', 26", 30, 31, 33, 34).

5. The measuring device according to any one of claims 1 to 4, characterized in that The pattern recognition operator recognizes event patterns and / or interference patterns.

6. The measuring device according to any one of claims 1 to 4, characterized in that The operator network comprises a free mode operator (9).

7. The measuring device according to any one of claims 1 to 4, characterized in that The operator network includes a notification operator, a machine control operator and / or a difference operator (35).

8. The measuring device according to any one of claims 1 to 4, characterized in that A plurality of operators are arranged in parallel or in series.

9. The measuring device according to any one of claims 1 to 4, characterized in that A plurality of mode operators are arranged in parallel or in series.

10. A mobile device, characterized in that: The device is designed as a computing unit with an input data stream and / or an output data stream, which is intended to be integrated into a measuring device according to one of claims 1 to 9 .

11. The mobile device according to claim 10, characterized in that The mobile device is designed as an electrically autonomous device.

12. The mobile device according to claim 10 or 11, characterized in that The mobile device is designed as a sensor or a display module.

13. A method for identifying events in a material processing process and / or a material manufacturing process based on a multidimensional data stream with time-resolved frequency information and energy information obtained during the process using interference patterns and / or event patterns and / or energy data, characterized in that The data pass through a mode calculator or a plurality of mode calculators arranged in parallel or in series and are supplied to a decision calculator (6) or a plurality of decision calculators (6) arranged in parallel or in series of a measuring device according to one of claims 1 to 9.

14. A device for identifying events in a material processing process and / or a material manufacturing process based on a multidimensional data stream with time-resolved frequency information and energy information obtained during the process using interference patterns and / or event patterns and / or energy data, characterized in that The data pass through a mode calculator or a plurality of mode calculators arranged in parallel or in series and are supplied to a decision calculator (6) or a plurality of decision calculators (6) arranged in parallel or in series of a measuring device according to one of claims 1 to 9.

15. A graphical user interface for constructing and / or editing a computing unit network of a measuring device according to any one of claims 1 to 9, characterized in that Graphical location and associatability of predefined computing elements of the measuring device.

16. A measuring device comprising a graphical user interface according to claim 15. 17 . A device for carrying out the method according to claim 13 , comprising a computing network of a measuring device according to one of claims 1 to 9 and / or comprising a graphical user interface according to claim 15 .

Images