DE102019215268A1 - Method and device for operating an electrochemical machining system - Google Patents

Abstract

The invention relates to a method for operating an electrochemical machining process, with the following steps:
- Determination (S2, S4, S5, S6) of process parameters for operating an electrochemical machining system with the aid of a Bayesian optimization method, the Bayesian optimization method being based on a quality function that can be trained with operating point data and that is successively refined, the operating point data being operating points which Represent process parameters for the electrochemical machining process, and assign setting parameters to a quality measure, the method iteratively applying a predetermined acquisition function to improve the quality function;
- Carrying out (S7) the electrochemical machining process depending on the trained quality function.

Description

Technical area

The invention relates to electrochemical machining systems. The invention also relates to measures for controlling an electrochemical machining process.

Technical background

Electrochemical machining systems use an electrolytic process to machine metallic materials of great hardness. For this purpose, a tool electrode is positioned with a gap distance over a surface of a workpiece in an electrolytic process. An electrolyte is arranged between the tool and the workpiece and a DC voltage is applied across the tool electrode and the workpiece, so that the tool electrode serves as the cathode and the workpiece as the anode. The current flow through the electrolyte leads to a process of dissolution of the material of the workpiece, so that material is removed from the workpiece.

Due to the material removal process, the tool electrode must be adjusted to maintain an effective gap size. The process of material removal accelerates as the gap size decreases, but too small a gap size can increase the risk that the workpiece or the tool electrode will be damaged due to a short-circuit current.

To control such a machining system, the advance of the tool and the DC voltage are usually determined heuristically. The control is only adapted from time to time if the structure of the workpiece changes or if there are influences due to the material quality. These adjustments are time-consuming and lead to long set-up times for the machining system and thus to high production costs.

Previous alternative approaches provide, for example, model-based controls that regulate a constant gap size, but this leads to unsatisfactory results due to the fact that the material removal process cannot be modeled. In particular, the setpoint curves must be specified heuristically, such as off EP2868417 is known.

Disclosure of the invention

According to the invention, a method for parameterizing an electrochemical machining system according to claim 1 and a corresponding device and an electrochemical machining system according to the independent claims are provided.

Further refinements are given in the dependent claims.

• - Bestimmen von Prozessparametern zum Betreiben eines elektrochemischen Bearbeitungssystems mithilfe eines Bayes'schen Optimierungsverfahren, wobei das Bayes'sche Optimierungsverfahren auf einer mit Betriebspunktdaten trainierbaren Qualitätsfunktion basiert, die sukzessive präzisiert wird, wobei die Betriebspunktdaten Betriebspunkte, die Prozessparameter für den elektrochemischen Bearbeitungsprozess darstellen, und Einstellungsparameter einem Qualitätsmaß zuordnen, wobei das Verfahren iterativ zum Verbessern der Qualitätsfunktion eine Aquisitionsfunktion anwendet;
• - Durchführen des elektrochemischen Bearbeitungsprozess abhängig von der trainierten Qualitätsfunktion.

According to a first aspect, a method for operating an electrochemical machining process is provided, with the following steps:

• - Determination of process parameters for operating an electrochemical machining system with the aid of a Bayesian optimization method, the Bayesian optimization method being based on a quality function that can be trained with operating point data and that is successively refined, the operating point data representing operating points, the process parameters for the electrochemical machining process, and setting parameters assign a quality measure, the method iteratively applying an acquisition function to improve the quality function;
• - Carrying out the electrochemical machining process depending on the trained quality function.

The reduction in the size of the gap between the tool electrode and the workpiece is directly related to the faster material removal and therefore to the reduction of production costs. However, if the tool electrode touches the workpiece, the resulting short-circuit current leads to immediate damage or destruction of the tool electrode and / or the workpiece. Thus, a suitable parameterization of the machining system, which tracks the tool electrode to the gap that increases due to material removal, must provide a compromise between reducing the gap size and adequate protection of the tool electrode and workpiece.

By using a Bayesian optimization algorithm, the above method automatically learns the existing safety limit while the method is being carried out, so that a short-circuit current between the tool electrode and the workpiece can only occur with a low, predeterminable probability. The gap size is optimized by minimizing it, taking safety criteria into account and based on process parameters such as the feed rate and the applied voltage, as well as setting parameters such as the material parameters of the workpiece to be machined, the machining tool and the electrolyte.

By optimizing the process parameters for machining using an electrochemical Machining method, the machining time can be reduced considerably without endangering the machining system and / or the workpiece.

Provision can be made for the method to iteratively apply the acquisition function and also a secondary condition to improve the quality function, the secondary condition defining a safety range which, as the next operating point, only allows one operating point at which a maximum machining current through the machining current is only possible with a predetermined probability is exceeded.

In particular, the safety range can be determined as a function of a predefined safety limit and specify ranges of the operating point data at which a short-circuit current can be avoided with the probability specified by the safety limit α.

Alternatively, the security area can be determined as a function of a separately trained security model, which in particular is also designed as a Gaussian process model.

It can be provided that the process parameters for the electrochemical machining process include a machining voltage and a drive.

According to one embodiment, the setting parameters can include one or more of the following parameters: material parameters of the workpiece and tool electrode, electrolyte parameters and one or more items of information about the contour of the workpiece.

Furthermore, the quality function can provide a machining flow specification as a quality measure, which specifies the machining flow directly or indirectly as an efficiency factor for a modeled machining flow.

• - Bestimmen von Prozessparametern zum Betreiben eines elektrochemischen Bearbeitungssystems mithilfe eines Bayes'schen Optimierungsverfahrens, wobei das Bayes'sche Optimierungsverfahren auf einer mit Betriebspunktdaten trainierbaren Qualitätsfunktion basiert, die sukzessive präzisiert wird, wobei die Betriebspunktdaten Betriebspunkte, die Prozessparameter für den elektrochemischen Bearbeitungsprozess darstellen, und Einstellungsparameter einem Qualitätsmaß zuordnen, wobei das Verfahren iterativ zum Verbessern der Qualitätsfunktion eine vorgegebene Aquisitionsfunktion anwendet;
• - Durchführen des elektrochemischen Bearbeitungsprozess abhängig von der trainierten Qualitätsfunktion.

According to a further aspect, a device for operating an electrochemical machining process is provided, the device being designed to:

• - Determination of process parameters for operating an electrochemical machining system with the aid of a Bayesian optimization method, the Bayesian optimization method being based on a quality function that can be trained with operating point data and that is successively refined, the operating point data representing operating points, the process parameters for the electrochemical machining process, and setting parameters assign a quality measure, the method iteratively applying a predetermined acquisition function to improve the quality function;
• - Carrying out the electrochemical machining process depending on the trained quality function.

According to a further aspect, there is an electrochemical machining system with a tool electrode, which is arranged in an electrolyte in order to electrochemically machine a workpiece, with a current or voltage source which is designed to apply a machining voltage to the electrolyte, and with the above device intended.

Figure list

• 1 eine schematische Darstellung eines elektrochemischen Bearbeitungssystems zur Bearbeitung eines Werkstücks; und
• 2 ein Flussdiagramm zur Veranschaulichung des Verfahrens zum Betreiben des elektrochemischen Bearbeitungssystems.

Embodiments are explained in more detail below with reference to the accompanying drawings. Show it:

• 1 a schematic representation of an electrochemical machining system for machining a workpiece; and
• 2 a flow chart illustrating the method for operating the electrochemical processing system.

Description of embodiments

1 shows a schematic representation of an electrochemical processing system 1 . The editing system 1 includes a workpiece holder 2 for holding a workpiece 3 from which material is to be removed. The workpiece 3 is formed from a conductive material, in particular a metallic material, and is in electrically conductive connection with the workpiece holder 2 .

It's a tool electrode 4th provided at a distance from the surface of the workpiece 3 is arranged. Thus, there is between the opposing surfaces of the tool electrode 4th and the workpiece 3 a gap is provided with a gap distance s. The workpiece holder 2 , the workpiece 3 and the tool electrode 4th are in an electrolyte 5 arranged. Between the workpiece holder 2 and the tool electrode is a machining voltage U by a voltage source 6th created so that when operating the machining system 1 with the tool electrode 4th as the cathode and the workpiece 3 as the anode, the current flowing through the electrolyte causes the workpiece to dissolve 3 leads and the workpiece 3 according to the shape of the tool electrode 4th is shaped. The dissolution process accelerates when the gap size s is reduced, whereby a short circuit occurs when the gap size is too small can that of the workpiece 3 and / or the tool electrode 4th due to an occurring short-circuit current.

During the process of forming the workpiece 3 must be the tool electrode 4th be tracked in order to keep the enlargement of the gap distance s due to the material removal constant. This is the tool electrode 4th via a feed device 7th movable, so that the machining process is controlled by adjusting the gap distance and the amount of material removed by the advance of the tool electrode 4th can be specified.

The machining process is carried out using a control unit 10 controlled. The machining process is controlled according to a function block diagram, as shown in 2 is shown. The gap size becomes the gap between the tool electrode 4th and the surface of the workpiece 3 set. For this purpose, the gap size is specified as a manipulated variable and can be adjusted via the current through the electrolyte, since the gap size itself cannot be measured.

The tool electrode 4th is with a predefined defined feed rate f in the direction of the workpiece 3 emotional. For the model-based control, the gap size s is modeled as resistance, that of the conductivity K of the electrolyte 5 and the area A of the electrode depends. In addition, the ohmic resistance of the interface between the tool electrode 4th and the electrolyte 5 referred to as interface resistor R. Furthermore, it is taken into account that before a current flow through the electrolyte occurs, the applied machining span must exceed _{a threshold potential U 0.} With this, the electric current I can be calculated from an equivalent circuit model as follows: $$\mathrm{I.}=\frac{U-U_0}{\mathrm{R.}+\frac{\mathrm{S.}}{k\mathrm{A.}}}$$

According to Faraday's law, the volume of the dissolved material is directly proportional to the electric current. The corresponding efficiency factor η relates to the specific material and the electrolyte properties as well as the contour of the workpiece to be machined 3 . The resulting change in gap size ṡ can be calculated as: $$\dot{s}=n\mathrm{I.}-f$$

In the control unit 10 a method, such as an optimal control or regulation, is implemented that specifies the process parameters, namely the machining voltage U and the feed rate f, based on a measured machining current I. The machining voltage U and the feed rate f are also specified based on a specified gap size or nominal current and based on setting parameters such as material parameters of the workpiece and tool electrode, electrolyte parameters and information about the contour of the workpiece. If, for example, a soft workpiece is used, the feed rate and the voltage should be set correspondingly smaller than with a hard workpiece. The values for feed and tension must also be specified conservatively if a strongly curved or uneven workpiece is used. There it can happen that the flow speed of the electrolyte increases at narrow points, which in turn favors faster material removal locally.

The processing system can be adapted to a given workpiece using the method described below.

To determine which is in the control unit 10 implemented a function which, in conjunction with the flowchart of the 2 is described in more detail. The method can be implemented in a data processing device as a software algorithm and / or hardware algorithm.

In step S1 are initially permissible and safe operating points of the machining system 1 given. Operating point data assign operating points, such as the machining voltage U and the feed rate f, to a corresponding gap size or a corresponding machining current I.

A safety limit α is also specified. The safety limit α indicates the probability with which the occurrence of a short circuit should be avoided.

A Bayesian optimization process is now started.

Bayesian optimization methods for the determination of optimized process parameters apply test process parameters iteratively and optimize them in an efficient way. A quality function is modeled using a Gaussian process regression in order to model the performance of the electrochemical machining process as a function of its process parameters, the Gaussian process model based on the result of the test procedure on a physical variable determined with the respective test process parameter is created. The quality function inputs Quality measure depending on process parameters of the electrochemical machining process and setting parameters.

Basically, the problem is to find process parameters (operating points) that lead to an optimized machining voltage and an optimized feed rate, i.e. an electrochemical machining process with rapid material removal while avoiding a short circuit. For this purpose, a quality function (cost function) that is dependent on the process parameters of the electrochemical machining process is evaluated.

In general, Bayesian optimization is used when uncertainty (e.g. due to measurement noise) is to be taken into account. This is also suitable if an unknown function is to be optimized for a so-called "black box" function. This unknown function f can only be evaluated and observed (possibly affected by noise) for a value x. The observed value y results as y = f (x) + e, where e denotes the noise.

In addition, it is assumed that every evaluation or measurement of the unknown function f is “expensive”, ie it causes costs, in the sense that the implementation of a test method for measuring the unknown function causes a certain, in particular high effort, as is the case with a Execution of a test procedure in the test facility is the case. Due to the "expensive" measurement of the unknown function, it is desirable that only a few measurements have to be carried out for the optimization or that the costs for the measurements (especially determined by the time and material expenditure) are as low as possible.

Under certain assumptions, such as the continuity of the unknown function, the quality function can be approximated with a data-based function model, such as a Gauss process regression. A Gaussian process is a universal function approximator that is used as a surrogate function for the unknown quality function. In general, Gaussian processes are understood to mean temporal, spatial or any other functions whose function values can only be modeled with probabilities due to incomplete information. With the help of functions of the expected values, variances and covariances, a Gaussian process describes the function values as a continuum of correlated random variables in the form of a high-dimensional normal distribution. A sample of these results in a random function with certain desired properties.

Usually numerical optimization algorithms are based on a large number of very cheap evaluations / measurements of the optimization functions. However, if the function evaluations are complex, such as If, for example, one of the above test procedures is carried out, the conventional optimization algorithms can no longer be used. Instead, Gaussian processes are used, which include a modeling of the quality function in the form of a surrogate function. The quality function describes the behavior of the system and specifies a quality value depending on the parameters with which the system is operated.

For this purpose, after measuring the quality function at several evaluation points, i.e. test process parameters (test operating points) and observing the corresponding function values, i.e. the respective quality value (depending on the test method, the maximum temperature cycles, the maximum tensile force or the maximum shear force), using the Gaussian A model of the quality function can be established. One property of the Gaussian process is that the model prediction is very good in areas around the measured test process parameters and the quality function is approximated well. This is reflected in a low level of uncertainty in the functional model. Far away from evaluation points, the model predictions about the quality function become poor and the uncertainty increases with increasing distance from the measured test process parameters.

step S2 provides the training of a Gaussian process model from the operating point data provided, which maps the operating point data and the setting parameters to a machining current specification. The machining current specification can correspond to a machining current I directly or to the efficiency factor η, which can be used as a factor to apply the machining current I modeled according to the above model, which is determined according to the above formula from a last applied machining voltage. The machining current specification corresponds in the sense of the Bayesian optimization method to the quality measure, since this determines the speed of the material removal.

In step S3 a safety area S (α) is determined depending on the safety limit α. The safety area S (α) specifies areas of the operating point data at which a short-circuit current can be avoided with the probability specified by the safety limit α. The safety range can thus be specified as a function of the variance or uncertainty of the expected values of the trained Gaussian process model. In particular, the security area with an assumed Lipschitz continuity of the security function according to Srinivas et al., “Gaussian process opimization in the Bandit setting: No Regret and Experimental Design ", Arxiv 2010, Arxiv: 0912.3995v4.

Lipschitz continuity can be used to estimate how the safety function will develop in the worst case when leaving the known area. This makes it possible to estimate how far the operating point data may exceed the limit of the known safety range in order to still be safe (i.e. within the safety range) or permissible. In the case of Gaussian processes, this statement is of course probabilistic, i.e. it indicates a probability with which one is at least certain.

Alternatively, the security area S (α) can be determined in accordance with a separately trained security model, which can also be designed as a Gaussian process model, in accordance with one in the publications Werkenkamp et al: "Bayesian Optimization with safety constraints: safe and automatic parameter tuning in robotics", Arxiv: 1602.04450 V1, and Schreiter et al .: "Self-exploration for active learning with Gaussian processes", ICML 2015 described procedure.

A possible strategy to optimize the quality function is to evaluate the quality function in many different places (e.g. on a regular grid) and to assume the lowest observed function value as the result of the optimization. This procedure is inefficient and many measurement processes with the test procedures are necessary, with a correspondingly high level of effort, to find the optimum.

Instead of this approach, the Gaussian process is used to select new test process parameters. This is done in step S4 In each recursion, a new test process parameter for measuring the quality function is selected in such a way that, on the one hand, this improves the Gaussian process model, so that the uncertainty of the Gaussian process is reduced. For this purpose, the test process parameters are usually selected in areas in which the quality function has not yet been evaluated (exploration). On the other hand, the new test process parameters for measuring the quality function are selected in such a way that the goal of optimizing the quality function, ie minimizing or maximizing it, is achieved as quickly as possible or with the smallest possible number of measurements with the test process parameters. For this purpose, test process parameters are preferred that promise low function values based on the Gaussian process (exploitation).

These two opposing criteria are weighed up by what is known as an acquisition function. In the Bayesian optimization method, the measurements for determining the quality function are optimized with the aid of the acquisition function in such a way that they do not necessarily have the lowest uncertainty overall, but rather the greatest possible information about the position of the optimum, i.e. H. have those process parameters at which the highest possible quality value can be achieved.

The acquisition function uses parameters of the quality function, which is described by a Gaussian process model, such as the Gaussian process mean µ (x) and the Gaussian process standard deviation σ (x). An example is the so-called Lower Confidence Bound (LCB) acquisition function or the Upper Confidence Bound (UCB) acquisition function (depending on whether a high quality measure is better or worse), which are described as follows: LCB ( x) = µ (x) - kσ (x) or UCB (x) = µ (x) + kσ (x). In practice, the factor k is often set to a constant value, for example k = 2. This new criterion can be efficiently minimized or maximized using common gradient-based methods and the minimum of LCB (x) or the maximum from UCB (x) then forms the new test process parameters for the quality function. It should be noted here that an optimization domain must be defined for the optimization of the acquisition function, in which the next test process parameters are searched for. This domain is typically chosen on the basis of experience and / or expert knowledge.

In addition to the above acquisition functions, other acquisition functions are known, such as Expected Improvement (EI), Probability of Improvement (PI) or so-called entropy search methods that are based on information-theoretical considerations.

ermittelt, mit der Nebenbedingung, dass der Betriebspunkt sich innerhalb des Sicherheitsbereichs S(α) befindet, d.h. mit der durch die Sicherheitsgrenze vorgegebenen Wahrscheinlichkeit nicht zu einem Kurzschlussstrom führt.The acquisition function is used in step S4 a new operating point accordingly $$\begin{array}{c}x*=\underset{x}{{\displaystyle \text{argmax}}}\mu \left(x\right)+\beta \sigma \left(x\right)\\ s.t.xin\mathrm{S.}\left(\alpha \right)\end{array}$$

determined, with the secondary condition that the operating point is within the safety range S (α), ie does not lead to a short-circuit current with the probability specified by the safety limit.

The following is in step S5 the operating point found is specified as a new process parameter and the processing system is measured accordingly.

Measuring the machining system 1 takes place initially with the feed rate f and the machining voltage U specified by the operating point. By increasing either the feed rate and / or the machining voltage up to a limit at which the measured machining current rises significantly above a predetermined maximum machining current, the operating point data can be provided which a combination of the process parameters and the resulting machining current, which defines a limit range of the permissibility of machining.

Will in step S5 at the in step S4 If the newly determined operating point to be measured has been determined that the maximum machining current is exceeded, then by reducing the feed rate f and / or the machining voltage U, the just still valid operating point at which the maximum machining current is not exceeded is determined. These operating point data are then provided as the next operating point data.

In step S6 a termination criterion is checked, for example the length of time that is to be expended for the optimization of the quality function, or the number of iterations or a suitable convergence criterion. If it results from this that the optimization is to be ended, the method starts with step S7 otherwise continue with step S2 continued.

In step S7 becomes the electrochemical machining system 1 operated with a machining current that results directly from the quality function. In particular, the machining current specification can be determined based on the process parameters and setting parameters, depending on which the electrochemical machining system is used 1 is operated.

If a data-based security model is used to determine the security area, this can be done in accordance with step S5 specific operating point data, which has led to the machining current exceeding the maximum machining current, can be trained. The safety model can be a classification model that assigns the operating points and the material parameters of the workpiece and tool electrode, electrolyte parameters and information about the contour of the workpiece to permissible or impermissible operating points. This makes it possible to determine the safety areas according to the given safety probability.

QUOTES INCLUDED IN THE DESCRIPTION

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

Patent literature cited

• EP 2868417 [0005]

Non-patent literature cited

• Werkenkamp et al: "Bayesian Optimization with safety constraints: safe and automatic parameter tuning in robotics", Arxiv: 1602.04450 V1, and Schreiter et al .: "Self-exploration for active learning with Gaussian processes", ICML 2015 [0043]

Claims (11)

Computer-implemented method for operating an electrochemical machining process, with the following steps: - Determination (S2, S4, S5, S6) of process parameters for operating an electrochemical machining system with the aid of a Bayesian optimization method, the Bayesian optimization method being based on a quality function that can be trained with operating point data and that is successively refined, the operating point data being operating points which Represent process parameters for the electrochemical machining process, and assign setting parameters to a quality measure, the method iteratively applying a predetermined acquisition function to improve the quality function; - Carrying out (S7) the electrochemical machining process depending on the trained quality function. Procedure according to Claim 1 , the method iteratively applying the acquisition function and an additional constraint to improve the quality function, the constraint defining a safety area (S (a)) (S3) which, as the next operating point, allows only one operating point at which a maximum machining current is passed through the machining current is exceeded only with a given probability. Procedure according to Claim 2 , the safety range (S (a)) being determined as a function of a predetermined safety limit (a) and indicating ranges of the operating point data at which a short-circuit current can be avoided with the probability specified by the safety limit (a). Procedure according to Claim 2 , the security area (S (a)) being determined as a function of a separately trained security model, which in particular is also designed as a Gaussian process model. Method according to one of the Claims 1 to 4th , wherein the process parameters for the electrochemical machining process include a machining voltage and a drive. Method according to one of the Claims 1 to 5 , wherein the setting parameters include one or more of the following parameters: material parameters of the workpiece and tool electrode, electrolyte parameters and one or more details about the contour of the workpiece. Method according to one of the Claims 1 to 6th , wherein the quality function provides a machining flow specification as a quality measure, which specifies the machining flow directly or indirectly as an efficiency factor for a modeled machining flow. Device for operating an electrochemical machining process, the device being designed for: - Determination of process parameters for operating an electrochemical machining system with the aid of a Bayesian optimization method, the Bayesian optimization method being based on a quality function that can be trained with operating point data and that is successively refined, the operating point data representing operating points, the process parameters for the electrochemical machining process, and setting parameters assign a quality measure, the method iteratively applying a predetermined acquisition function to improve the quality function; - Carrying out the electrochemical machining process depending on the trained quality function. Electrochemical machining system (1) with a tool electrode (4) which is arranged in an electrolyte (5) in order to electrochemically machine a workpiece, with a current or voltage source (6) which is designed to supply a machining voltage (U) to apply the electrolyte (6), and with a device after Claim 8 . Computer program with program code means which is set up to implement a method according to one of the Claims 1 to 7th execute when the computer program is executed on a computing unit. Machine-readable storage medium with a computer program stored thereon Claim 10 .

Images