Method and device for identifying process status of plasma arc method

Abstract

The invention relates to a method for identifying a process state of a plasma arc method, in which a workpiece (1) is treated by a plasma torch (2), measured values of a time curve of physical variables of the plasma arc method are detected, and a signal curve is determined from the measured values (S3-1). The signal curve is pattern-recognized (S3-2) by extracting at least one feature from the signal curve and classifying the feature, and the signal curve is assigned to a specific state of the plasma arc method according to the classification of the feature (S3-4).

Description

Technical Field

[0001] The invention relates to a method and a device for identifying process states of a plasma arc method. Background Art

[0002] In plasma arc methods, such as plasma cutting, events may occur which have a negative impact on the result to be achieved. This concerns, for example, the quality of the cut, damage to the plasma torch or the cut material, in particular due to unforeseen or unforeseen malfunctions or the use of consumable parts outside the recommended operating times. In order to counteract this, plasma arc methods are monitored with the aim of being able to detect such events in good time and, if necessary, to abort the process or change one or more process parameters. This not only helps to optimize the respective plasma arc method, but also protects the plasma torch or the workpiece used for this purpose.

[0003] Several methods are known from the prior art. Typically, a single parameter, such as the amplitude or rate of change of a voltage, is determined and, if a threshold value is exceeded, corresponding measures are initiated. For example, US 5,750,957 A discloses a method in which the standard deviation of an electrical signal is monitored and used to assess the quality of a process.

[0004] However, a disadvantage of this approach is that the parameters used for quality control usually depend on method parameters (such as the speed of the plasma arc method and the material thickness of the workpiece), which means that the threshold value to be determined requires knowledge of these cutting parameters and the approach is difficult to generalize.

[0005] The penetration of the workpiece as cutting material may also manifest itself in different signal curves at different material thicknesses, so that a simple comparison with the threshold value usually only gives inadequate results. The same applies to consumable failures, which can lead to different signal patterns depending on the cathode used, for example. Summary of the invention

[0006] The invention therefore solves this problem by specifying a method and a device which avoid the above-mentioned disadvantages and by which process states of a plasma arc method, in particular fault situations, can be determined quickly and reliably.

[0007] This problem is solved according to the invention by a method according to claim 1 and a device according to the corresponding claim. Advantageous embodiments and developments are described in the dependent claims.

[0008] In a method for identifying a plasma arc method or a process state of a plasma arc method, a workpiece is treated by a plasma torch and measured values ​​of a time curve of a physical variable of the plasma arc method are detected. A signal curve is determined from the measured values ​​and a pattern recognition is performed on the signal curve. At least one feature is extracted from the signal curve and classified. Finally, the signal curve is assigned to a specific state of the plasma arc method based on the classification of the feature.

[0009] By recording the time curve of a physical variable associated with a plasma arc method, a (time) signal curve of the physical variable is determined, which is a curve that is specific to the corresponding method. By not only looking for situations where a threshold value is exceeded (which may also trigger false alarms due to strong noise, deviations from process parameters, etc.), but by using pattern recognition to find patterns similar to known process states, the process state can be determined more reliably than known methods. For this purpose, features are obtained from the signal sequence, which can be assigned to the patterns to be detected and classified into the same category (classification). This classification, i.e. the assignment of features to specific states of the plasma arc method, allows a more reliable determination of the state of the plasma arc method than a simple threshold analysis. Therefore, the method can be used in particular to monitor arc stability, detect failures of consumables and stop operation to protect the plasma torch from irreparable damage. In addition, situations where the cathode service life is shortened can be determined, such as when the arc burns on the hole or passes over the end of the workpiece after the perforation process, so that measures can be taken to maintain the cathode service life by adjusting process parameters (such as reducing the plasma torch height or reducing the cutting current or cutting gas pressure).

[0010] The signal curve can be obtained directly in the time domain or also in the frequency domain, for example by converting the time curve of the measured value into the frequency domain using a Fourier transformation.

[0011] It can be provided that the at least one physical variable is a voltage, a voltage drop and / or a current. These variables are usually relevant to the plasma arc and can be reliably determined.

[0012] The at least one feature can be determined from the amplitude and / or the amplitude spectrum of the measured value, wherein the feature is preferably determined from the sum of the amplitude spectra in at least one predetermined frequency range, particularly preferably in at least two predetermined frequency ranges. In particular, when certain relevant frequency ranges of the amplitude spectrum are used, process states in which frequencies in these frequency ranges frequently occur can be determined with high accuracy. The term "signal curve" can therefore describe both the time curve of the measured value and the amplitude spectrum obtained therefrom.

[0013] However, the at least one characteristic can also be determined from a parameterization of at least one or more parts of the entire signal curve or from a parameterization of the entire signal curve. This enables in particular a fast and reliable detection of process states which can be detected by changes over time. Preferably, the characteristic is determined by parameterization of one or more individual parts of the signal curve by forming a series of mean values ​​and / or rates of change and / or regression values ​​of a regression analysis.

[0014] The number of recorded measured values ​​forming the signal curve can be at least 500 measured values, preferably 10,000 measured values, particularly preferably 15,000 measured values, in order to achieve a sufficiently high accuracy and thus reliability of the method.

[0015] The time between two individual measured values ​​forming the signal curve is a maximum of 1 ms, preferably 0.05 ms, particularly preferably 0.03 ms, in order to be able to determine the process state quickly enough. The pattern recognition is usually performed in real time, i.e. within a maximum of 1.5 seconds, or is provided in the form of a lookup table in which the result of the pattern recognition is calculated for any feature combination and stored in the table.

[0016] The time for measuring and forming the signal curve can be up to 1500 ms, preferably up to 500 ms, particularly preferably up to 350 ms, in order to obtain information about the process state in real time. The measuring time is preferably longer than the time required to obtain a feature or recognition pattern.

[0017] In order to increase the reliability of the method, it is usually also intended to determine at least two features and use them to determine the process state, that is, to extract at least two features from the signal curve and classify the features. However, the number of features is usually smaller than the number of individual measured values.

[0018] The classification categories typically include at least one stable process state and / or at least one unstable process state. If a stable process state exists, the plasma arc method continues unchanged, while if an unstable process state exists, the plasma arc method is preferably interrupted to avoid damage. An unstable process state may be caused by or indicate a damaged or worn consumable part. Other states may also be classified that indicate a fault (e.g., hafnium ejection from the cathode, or double arcing between the cathode and the nozzle), but do not require any intervention in the process.

[0019] A plasma arc method is usually plasma cutting using a plasma cutting torch. However, in principle, all processes for treating materials using a plasma arc can be considered as plasma arc methods. In this case, the categories can include idling, pilot arc ignition, pilot arc burning, workpiece perforation, penetration of the workpiece, cutting of the workpiece, crossing the cut, crossing the edge of the workpiece, the presence of damaged consumables and / or the presence of worn consumables or multiple worn consumables as process states. Alternatively or additionally, the wear of the consumables can be divided based on the degree of burnout of the cathode (measured in millimeters) and / or based on the degree of burnout relative to a specified service life, or in particular the classification into the categories of the presence of damaged consumables, the presence of new consumables, the presence of old consumables and / or worn consumables. For example, in the first case, the classification can be made based on the remaining length of the cathode, i.e., for example, a length in the range of 0 mm to 1 mm, a length in the range of 1 mm to 1.8 mm, a remaining length greater than 1.8 mm. With regard to the maximum service life, which manifests itself as the maximum possible burn-in, it is also possible to classify using percentages between 0% (ie no burn-in) and 100% (ie maximum burn-in and reaching the maximum service life).

[0020] Depending on the result of the classification, an optical, acoustic, electrical and / or electronic signal can be generated or output, or a signal that can be evaluated optically, acoustically, electrically and / or electronically can be generated or output to stabilize or terminate the plasma arc process. The signal can also be a warning signal prompting human intervention.

[0021] A device for identifying a process state of a plasma arc method has a plasma torch which is arranged to process a workpiece. The device also includes a measured value recording device which is configured to record measured values ​​of a time curve of a physical variable of the plasma arc method and to determine a signal curve based on the measured values. Furthermore, an evaluation unit is part of the device and is configured to extract at least one feature from the signal curve and to classify the feature and to assign the signal curve to a specific state of the plasma arc method based on the classification of the feature.

[0022] The plasma torch generally has at least one electrode and a nozzle and is powered by electrical energy from a current source or a voltage source, which can also be part of the device. The workpiece serves as a second electrode. Furthermore, a control unit can be provided, which controls the current source or the voltage source, in particular also based on signals received from the evaluation unit.

[0023] The device may also have an inductive component electrically connected in series with the plasma torch. In this case, the measured value recording device is also configured to measure the current flowing through the inductive component or the voltage drop across the inductive component. At least part of the current that also flows through the electrode of the plasma torch typically flows through the inductive component.

[0024] The voltage to be measured may be the voltage between an electrode of a plasma torch and a workpiece, the voltage between the electrode and a nozzle, and / or the voltage drop across an inductive component (usually configured as a choke).

[0025] The current to be measured can in particular be the current flowing from the electrode to the nozzle and / or from the electrode to the workpiece.

[0026] The described device is generally configured to carry out the described method, ie the explained method can be carried out by means of the device.

[0027] A computer program product comprises a computer program having software components for carrying out the described method and / or controlling the described device when the computer program is executed in an automation system or a computing unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Exemplary embodiments of the present invention are shown in the accompanying drawings and will be referred to below. Figure 1-Figure 11 Provide explanation.

[0029] In the figure:

[0030] Figure 1 shows a schematic side view of an apparatus for plasma arc processing;

[0031] Figure 2 a graph showing measured values ​​of a physical variable recorded over a period of time;

[0032] Figure 3 A schematic method sequence for pattern recognition is shown;

[0033] Figure 4 Shown with Figure 3 A view of the detailed method sequence of the corresponding pattern recognition;

[0034] Figure 5 A schematic representation of a signal curve consisting of voltage and current for determining the condition of a wearing part is shown;

[0035] Figure 6 shows a schematic representation of a signal curve in the case of a defective wearing part;

[0036] Figure 7 shows that when crossing a seam or leaving the board, Figure 6 Schematic representation of the corresponding signal curve;

[0037] Figure 8 A schematic representation of the summation of the amplitude spectra is shown;

[0038] Fig. 9 shows a schematic graph of pattern recognition using parameterization;

[0039] Fig.10 A schematic representation of the feature space is shown; and

[0040] Fig.11 A true representation of the feature space is shown. DETAILED DESCRIPTION

[0041] Figure 1 A schematic side view of a device for carrying out and monitoring a plasma arc method is shown in the illustrated exemplary embodiment of plasma cutting. A plasma torch 2 or plasma cutting torch having a nozzle and an electrode (and other components, such as a coolant inlet, etc., which are known in the prior art and are therefore not listed here for the sake of clarity) is electrically connected to a power supply 4 and emits a plasma arc 3 for cutting a workpiece 1, which is also electrically connected to the power supply 4. In the illustrated exemplary embodiment, a choke 7 is electrically connected in series with the plasma torch 2 as an inductive component, but the choke 7 can also be omitted in other embodiments. The measured value recording device 5 serves to record measured values ​​of the time curve of a physical variable, such as a current, a voltage or a voltage drop, in particular during the cutting process, and to evaluate the measured values ​​in an evaluation unit 6.

[0042] The evaluation unit 6 is configured to perform pattern recognition on the signal curve based on the measured values ​​obtained and the signal curve determined by the measured value recording device 5. At least one feature, but usually two features, are extracted and classified by the evaluation unit 6 from the signal curve, wherein different classification categories represent specific states of the plasma arc process. Depending on the determined category, the evaluation unit 6 can output an electrical signal to the current source 4, which controls the current accordingly. In other exemplary embodiments, a voltage source can also be used instead of the current source 4, which controls the voltage accordingly.

[0043] exist Figure 1 In the exemplary embodiment shown, the measured value recording device 5 is also configured to measure the voltage drop across the choke 7. The measured voltage may be the voltage between the electrode and the workpiece 1, or the voltage between the electrode and the nozzle, and / or the voltage drop across the choke 7. The measured current may be the current flowing from the electrode to the nozzle and / or the current flowing from the electrode to the workpiece 1.

[0044] Typically, the evaluation unit 6 and the measured value recording device 5 are configured as a single computing unit, on which a computer program product can typically be run. The computer program product has software components which control the computer program product when it is run in the computing unit as an automation unit. Figure 1 The apparatus shown may perform a method described in more detail below.

[0045] Figure 2A graph is shown, in which a typical curve of a variable (such as voltage or current) recorded by a measured value recording device 5 is shown as a function of time (in seconds). The graph shows the cutting current in amperes (top curve), the voltage between the electrode and the workpiece 1 in volts (middle curve), and the voltage between the electrode and the nozzle in volts. In the range from 1.5 seconds, the fluctuation increases, and in the range from 2.9 seconds, the voltage drops sharply. Any of these deviations indicates that there is an error in the process and that the cutting process needs to be stopped quickly to protect the plasma torch. Among other things, fluctuations in the range of 0.7 seconds are caused by hafnium, etc. in the arc or double arc between the cathode and the nozzle. However, since a threshold value (which, according to the prior art, is intended to be used as a fault indicator) may be exceeded (and the process may be interrupted), in particular at fluctuations at 0.7 seconds, and then the process may continue normally, pattern recognition is performed in the described method in order to more reliably detect actual fault events that require interruption of the plasma arc method.

[0046] Figure 3 A schematic diagram of a corresponding pattern recognition process is shown. In step S3-1, the signal curve determined by the measured value recording device 5 from the measured values ​​is stored and, if necessary, transmitted to the evaluation unit 6. In step S3-2, at least one feature is determined from the signal curve (or, if there are signal curves for multiple physical variables, from each signal curve or at least from at least two signal curves). In step S3-3, pattern recognition is performed by similarity comparison to determine which known mode or state of the plasma arc method the current signal curve is similar to.

[0047] Pattern recognition is understood here as the ability to recognize regularities, repetitions, similarities, or uniformities in a set of data (such as measurement values or signal curves). The features defined for the pattern are used to distinguish it from the content of other classes. In addition to the sensor-detected data (as described above), preprocessing for data reduction can also be provided to improve the data quality. By extracting features as part of pattern recognition, the pattern is thus transformed into a feature space during feature extraction. In the feature space, the pattern is represented as a point, where the features can usually be mathematically represented as vectors, namely the so-called feature vectors. These features are measured as numerical values and combined into (multidimensional) feature vectors. If necessary, an intermediate step of feature reduction can be included in the method sequence to confine the sample to the key elements. The classification operation assigns the feature vector to be inspected to the class that has the greatest correspondence or similarity to the feature vector to be inspected, that is, the features of this class have the greatest correspondence or similarity to the feature vector to be inspected. This can be done using a mathematical function (such as the scalar product between the feature vector to be analyzed and one or more vectors representing the corresponding class), or using a classifier that assigns features to different classes and thus maps the process state. For example, if a fault state of a vulnerable part is detected in a subsequent step S3-4 (by classifying the sample into the corresponding class), the power supply 4 can be switched off in step S3-5.

[0048] During preprocessing, the signal can be passed through, for example, a high-pass filter, a low-pass filter, or a band-pass filter. Alternatively or additionally, the measurement values can also be normalized. The actual features may vary and may include, for example, key figures of distribution functions, moments such as the expected value and variance, or correlations and convolutions.

[0049] In addition, a transform such as the discrete Fourier transform can also be used to transfer the measurement values into the frequency domain to obtain a more manageable feature space.

[0050] Figure 4 A detailed representation of pattern recognition is also shown in a flowchart. After starting in step S4-1, in step S4-2, a memory is created in the measurement value recording device 5 for a total of M samples that are sampling values. For the selected exemplary embodiment, the number of samples is at least 5000, and the time between recording two samples is at most 0.1 ms. In the example shown, the entire measurement period is at most 1 second.

[0051] In step S4-3, N samples of the voltage between the electrode and the nozzle are sampled and added to the memory (where N < M). Step S4-4 checks whether the memory is full, that is, whether there are a total of M samples in the memory. If not, the cutting process continues, and if so, the pattern recognition process is performed in steps S4-6, S4-7, and S4-8.

[0052] To this end, in step S4-6 of the exemplary embodiment shown, firstly M samples are preprocessed and features are obtained from the signal curve in step S4-7. Step S4-8 checks whether a predefined classifier provides the closest match to a specific class, for example the class of "cathode failure" or "consumable part failure". If not, the N oldest samples are deleted from the memory in step S4-5 and the process continues. If yes, the cutting process ends in step S4-9 and in step S4-10 an error message, for example "consumable part failure", is output on the display unit before reaching the end. In the exemplary embodiment shown, a "conventional process sequence" (where the plasma arc method continues) and an "unconventional process sequence" (where the process also continues, but a warning message is issued to the user, for example Figure 2 The fluctuation at 0.7 seconds shown in ) can be defined as other categories.

[0053] In general, a "teaching phase" can also be provided, in which the evaluation unit 6 stores a database with a plurality of patterns and associated categories, or teaches them using a classification procedure. For example, artificial neural networks or "support vector machines" can also be used here.

[0054] exist Figure 5 In the diagram, the voltage is plotted against the current to illustrate the different categories. Depending on the classification, the signal curve is divided into the categories "new consumables", "old consumables" or "worn consumables". Other categories can be generated based on cathode burnout in the range of 0% to 100% of the maximum possible burnout. Figure 6 and Figure 7 shows a signal curve in which the voltage (in V) is plotted against the time (in s) and the category "defective consumable" ( Figure 6 ), and the category "Breakout Board" ( Figure 7 ) and "across the seam" ( Figure 7 dotted line in the middle) or "Piercing process completed" ( Figure 7 Example signal curves (alternating dash and dot-dash lines in ). For example, categories such as "idle running", "igniting the pilot arc", "maintaining the pilot arc", "piercing the workpiece", "penetrating the workpiece", "cutting the workpiece" or "going over the workpiece edge" can also be defined and features can be assigned accordingly.

[0055] Figure 8The Fourier transform result of the signal curve obtained at a sampling rate of 50kHz is shown. In the frequency range shaded in gray, the amplitudes are added (Σ1 and Σ2) and the sum is used as a feature for classification, i.e. a feature vector can be formed from the obtained numbers and assigned to a specific class. By using the Fourier transform, the signal curve can be evaluated independently of the magnitude or similar parameters and the rate of change of the voltage amplitude. Figure 8 The signal curves for a conventional cutting process (solid curve) and for a fault in a consumable part (dashed line) are shown. There are beat frequencies between 50 Hz and 100 Hz, which are caused by the current source 4 and are therefore ignored in the evaluation. The two sums formed correspond to two features, and the resulting values ​​allow a clear distinction. In the exemplary embodiment shown, the data of the last 327 ms of the method are temporarily stored in a memory, resulting in 16384 samples, which can be reduced to only two key figures by determining the features in the amplitude spectrum to calculate the class assignment. In the embodiment shown, a "k nearest neighbor" classifier is used as a classifier. However, in other exemplary embodiments, depending on the intended application, a "linear support vector machine (linear SVM)", "radial basis function support vector machine (RBF SVM)", "Gaussian process", "random forest", "neural network", "adaptive boosting", "naive Bayes" or "quadratic discriminant analysis (QDA)" classifier can also be used, or the distance between the feature vectors associated with each class and the feature vector to be checked is calculated. The class assignment of the feature combination is calculated by determining the k nearest neighbor objects obtained from the classification in the training phase. Assign the feature or object to the category that appears most frequently among its k nearest neighbors.

[0056] Fig. 9 The parameterization of the individual sub-areas is shown. In the graphs (left for new consumables, right for consumables before the end of their service life), the voltage between the electrode and the nozzle is plotted against the current, with the ranges of 75A to 150A, 170A to 250A, and 250A to 300A being used for the regression calculations. The coefficients a1, b1, c1, a2, b2, c2 and a3, b3, c3, d3 obtained by the regression calculations can be used as features for the classification. However, in general, amplitudes, concave or convex curves and / or rising trends can also be used as features.

[0057] Fig.10 A schematic two-dimensional feature space is shown, where two different features are plotted on the axes and the feature vector formed by these features points to a specific point in the feature space. Fig.10 In the first step, features are collected from the training dataset ( Fig.10By training the classifier, a function can be obtained that automatically assigns categories to new feature vectors. Therefore, the classifier assigns the category with the "highest" match to the feature combination, as shown by the two arrows. Fig.10 In the exemplary embodiment shown, three categories are thus formed. Fig.11 An actual feature space is shown, where the endpoints of the feature vectors are plotted. The three categories correspond to “Consumable damage”, “Regular cutting area” and “Irregular process sequence”.

[0058] Features of the various embodiments disclosed only in the exemplary embodiments may be combined with one another and claimed separately.

Claims

1. A method for identifying a process state of a plasma arc method, wherein: A workpiece (1) is processed by a plasma torch (2), measured values ​​of a time curve of a physical variable of the plasma arc process are detected, and a signal curve is determined based on the measured values ​​(S3-1), wherein the signal curve is subjected to pattern recognition by extracting at least one feature from the signal curve and classifying the feature (S3-2), and the signal curve is assigned to a specific state of the plasma arc process based on the classification of the feature (S3-4).

2. The method according to claim 1, characterized in that The physical variable is voltage, voltage drop, and / or current.

3. The method according to any one of the preceding claims, characterized in that The at least one feature is determined from an amplitude and / or an amplitude spectrum of the measured values.

4. The method according to claim 3, characterized in that The at least one characteristic is determined based on the sum of amplitude spectra within at least one predetermined frequency range.

5. The method according to any one of the preceding claims, characterized in that The at least one characteristic is determined from a parameterization of the signal curve.

6. The method according to claim 5, characterized in that The at least one characteristic is determined by parameterizing the various portions of the signal curve by forming a series of mean values ​​and / or rates of change and / or regression values ​​of a regression analysis.

7. The method according to any one of the preceding claims, characterized in that At least two features are extracted from the signal curve and the at least two features are classified.

8. The method according to any one of the preceding claims, characterized in that The categories of the classification include at least one stable process state and at least one unstable process state.

9. The method according to claim 8, characterized in that In the case of classification as the stable process state, the plasma arc method continues unchanged, and in the case of classification as the unstable process state of the plasma arc method, the plasma arc method is terminated.

10. The method according to any one of the preceding claims, characterized in that The plasma arc method is plasma cutting and uses the following categories: idle, pilot arc ignition, pilot arc burning, workpiece penetration, through workpiece, cutting workpiece, over the cut, over the edge of the workpiece, damaged consumables present, new consumables present, old consumables present, and / or worn consumables present.

11. The method according to claim 10, characterized in that Based on cathode burnout or based on maximum service life, the categories are divided into the following categories: damaged consumables are present, new consumables are present, old consumables are present, and / or worn consumables are present.

12. An apparatus for identifying a process state of a plasma arc method, comprising: A plasma torch (2) for processing a workpiece (1), a measured value recording device (5) for recording measured values ​​of a time curve of a physical variable of the plasma arc method and determining a signal curve based on the measured values, an evaluation unit (6) for pattern recognition of the signal curve, the evaluation unit being configured to extract at least one feature from the signal curve and to classify the feature, and to assign the signal curve to a specific state of the plasma arc method based on the classification of the feature.

13. The device according to claim 12, characterized in that An inductive component (7) is connected in series with the plasma torch, wherein the measured value recording device (5) is configured to measure a current or a voltage drop as the physical variable. 14 . A computer program product having a computer program, the computer program comprising software components for executing the method according to claim 1 and / or controlling the device according to claim 12 when the computer program is executed in an automation system.

Images