CN116490314A - Processing condition searching device and processing condition searching method - Google Patents
Processing condition searching device and processing condition searching method Download PDFInfo
- Publication number
- CN116490314A CN116490314A CN202180079088.9A CN202180079088A CN116490314A CN 116490314 A CN116490314 A CN 116490314A CN 202180079088 A CN202180079088 A CN 202180079088A CN 116490314 A CN116490314 A CN 116490314A
- Authority
- CN
- China
- Prior art keywords
- machining
- machine learning
- unit
- parameter
- electric discharge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012545 processing Methods 0.000 title claims abstract description 121
- 238000000034 method Methods 0.000 title claims description 20
- 238000003754 machining Methods 0.000 claims abstract description 333
- 238000010801 machine learning Methods 0.000 claims abstract description 133
- 238000011156 evaluation Methods 0.000 claims description 53
- 238000004364 calculation method Methods 0.000 claims description 39
- 230000006870 function Effects 0.000 description 23
- 238000010586 diagram Methods 0.000 description 8
- 230000002787 reinforcement Effects 0.000 description 8
- 239000000463 material Substances 0.000 description 6
- 230000007423 decrease Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23H—WORKING OF METAL BY THE ACTION OF A HIGH CONCENTRATION OF ELECTRIC CURRENT ON A WORKPIECE USING AN ELECTRODE WHICH TAKES THE PLACE OF A TOOL; SUCH WORKING COMBINED WITH OTHER FORMS OF WORKING OF METAL
- B23H1/00—Electrical discharge machining, i.e. removing metal with a series of rapidly recurring electrical discharges between an electrode and a workpiece in the presence of a fluid dielectric
- B23H1/02—Electric circuits specially adapted therefor, e.g. power supply, control, preventing short circuits or other abnormal discharges
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23H—WORKING OF METAL BY THE ACTION OF A HIGH CONCENTRATION OF ELECTRIC CURRENT ON A WORKPIECE USING AN ELECTRODE WHICH TAKES THE PLACE OF A TOOL; SUCH WORKING COMBINED WITH OTHER FORMS OF WORKING OF METAL
- B23H7/00—Processes or apparatus applicable to both electrical discharge machining and electrochemical machining
- B23H7/14—Electric circuits specially adapted therefor, e.g. power supply
- B23H7/20—Electric circuits specially adapted therefor, e.g. power supply for programme-control, e.g. adaptive
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Automation & Control Theory (AREA)
- Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- Electrical Discharge Machining, Electrochemical Machining, And Combined Machining (AREA)
Abstract
The processing condition searching device (10) comprises: a machine learning unit (19) that learns machining conditions for electric discharge machining performed by an electric discharge machine (28), and searches for machining conditions output to the electric discharge machine (28) based on a learning result; and a machine learning parameter determination unit (15) that determines a machine learning parameter that represents a range of values of a control parameter that is set as a target of search by the machine learning unit (19) and a control parameter that is set as a target of search, among control parameters that constitute machining conditions, based on a machining specification specified in relation to the manner of electric discharge machining.
Description
Technical Field
The present invention relates to a machining condition search device and a machining condition search method for searching for machining conditions for electric discharge machining.
Background
An electric discharge machine that performs shape-carving electric discharge machining performs electric discharge machining by selecting appropriate machining conditions based on machining specifications such as a machining material, a shape of an electrode, or a machining depth. The electric discharge machine has difficulty in satisfactory machining unless the machining conditions are changed in accordance with the change in the machining state accompanying the progress of machining. In addition, when the processing speed is high or when high processing accuracy is required, the processing state may be changed due to a slight change in the state of the processing liquid or the state of the processing such as a quality difference of the processing materials for each lot of the processing materials. Therefore, the electric discharge machine needs to select not only the machining conditions from among the preset machining conditions, but also the machining conditions to be adjusted in accordance with the progress of machining or the change in the condition of performing machining.
Patent document 1 discloses an electric discharge machine that observes an electric discharge machining state during electric discharge machining and adjusts machining conditions based on the observation result. The electric discharge machine according to patent document 1 optimizes the value of the control parameter by observing the state of electric discharge machining while changing the value of the control parameter with respect to the control parameter constituting the machining condition.
Patent document 1: international publication No. 2019/202672
Disclosure of Invention
In the case of adjusting the machining conditions for electric discharge machining, it is necessary to adjust the values of the control parameters with respect to each of the plurality of control parameters. With respect to all the values that can be obtained for each of the plurality of control parameters, it is not easy to search for an optimal combination from among combinations of values of the respective control parameters. In the technique of patent document 1, since an optimum combination is searched for from among a large number of combinations related to the values of the respective control parameters, efficient searching of the processing conditions is difficult.
The present invention has been made in view of the above circumstances, and an object thereof is to provide a machining condition search device capable of efficiently searching for machining conditions for electric discharge machining.
In order to solve the above problems and achieve the object, a processing condition searching device according to the present invention includes: a machine learning unit that learns machining conditions for electric discharge machining performed by the electric discharge machine, and searches for machining conditions to be output to the electric discharge machine based on a learning result; and a machine learning parameter determination unit that determines a machine learning parameter indicating a range of values of a control parameter set as a target of search by the machine learning unit and a control parameter set as a target of search among control parameters constituting the machining conditions, based on the machining specification specified in relation to the electric discharge machining.
ADVANTAGEOUS EFFECTS OF INVENTION
The machining condition search device according to the present invention has an effect of enabling efficient search of machining conditions for electric discharge machining.
Drawings
Fig. 1 is a diagram showing a configuration of a processing condition search device according to embodiment 1.
Fig. 2 is a flowchart showing a procedure of processing performed by the processing condition search device according to embodiment 1.
Fig. 3 is a diagram showing an example of a determination rule used by the processing condition search device according to embodiment 1 to determine the machine learning parameter.
Fig. 4 is a diagram showing an example of a search range of processing conditions in the processing condition search device according to embodiment 1.
Fig. 5 is a flowchart showing a learning procedure performed by the learning unit of the processing condition search device according to embodiment 1.
Fig. 6 is a diagram showing a configuration example of hardware of the processing condition search device according to embodiment 1.
Detailed Description
The processing condition search device and the processing condition search method according to the embodiments will be described in detail below with reference to the drawings.
Embodiment 1.
Fig. 1 is a diagram showing a configuration of a processing condition search device 10 according to embodiment 1. The machining condition search device 10 searches for machining conditions for electric discharge machining performed by the electric discharge machine 28. The processing condition search device 10 searches for an optimal combination of values of a plurality of control parameters constituting the processing condition.
The machining condition search device 10 learns the machining conditions by machine learning during machining by the electric discharge machine 28, and searches for the machining conditions based on the learning result. The machining condition search device 10 learns machining conditions according to machine learning parameters described later.
The electric discharge machine 28 is a machine that performs shape-engraving electric discharge machining. The electric discharge machine 28 includes: an electrode; a driving device for controlling the distance between the electrode and the workpiece; and a machining power supply for generating an electric discharge between the electrode and the workpiece. The structure of the electric discharge machine 28 is not shown.
The processing condition search device 10 includes: a machining specification input unit 11 for inputting machining specification information; a machining condition creation unit 12 that creates a reference machining condition; a machine learning setting input unit 13 to which machine learning setting information is input; a determination rule storage unit 14 that stores a determination rule of the machine learning parameter; a machine learning parameter determination unit 15 that determines machine learning parameters; a machine learning unit 19; a machining condition adjustment result storage unit 20 that stores adjustment results of machining conditions; and a decision rule updating unit 21 that updates a decision rule of the machine learning parameter.
The machining specification indicates a manner of electric discharge machining of a machining material, a machining shape, a machining depth, important performance, or the like. The work material is a material of a work object. The machining shape is a shape of an electrode transferred to the workpiece. The machining depth is the depth of a portion to be machined in the workpiece. Important properties are important properties in processing, and are properties such as processing speed and processing stability. The user of the machining condition search device 10 inputs information of the machining specification to the machining specification input unit 11, thereby specifying the machining specification. The machining specification information is input from the machining specification input unit 11 to the machining condition creation unit 12 and the machine learning parameter determination unit 15.
The reference machining condition is a machining condition composed of initial values of control parameters, and indicates an initial condition at the start of search. The machining condition creation unit 12 creates a reference machining condition based on the machining specification. The machining condition creation unit 12 stores a plurality of representative machining conditions in advance. The machining condition creation unit 12 creates a reference machining condition by selecting a machining condition corresponding to the machining specification input unit 11 from among the plurality of stored machining conditions. The reference machining conditions are input from the machining condition creation unit 12 to the machine learning parameter determination unit 15 and the machine learning unit 19, respectively.
The machine learning setting information is information indicating a machining type of a search target set as a machining condition related to machine learning. The machining type is a machining type to be searched for in the machining conditions related to machine learning, and is a machining type such as rough machining or finish machining. The user inputs machine learning setting information to the machine learning setting input unit 13, thereby specifying a desired machining type. Information of the machine learning setting information is input from the machine learning setting input unit 13 to the machine learning parameter determination unit 15.
The machine learning setting information input to the machine learning setting input unit 13 may include user specification information for specifying a search target or a search range by a user. The user can specify the type of the control parameter to be searched or the range of the value of the control parameter to be searched by the user specification information. For example, in a case where a recipe has been determined with respect to a part of the processing conditions, the user can set an arbitrary limit in the search range of the processing conditions by inputting user specification information in accordance with the recipe. For example, when there is a control parameter that is not desired to be changed, the user can specify that the control parameter is excluded from the search target by inputting the machine learning setting information. The user can specify a combination of control parameters set as search targets by the user specification information. The input of the user-specified information is arbitrary.
The machine learning parameters include information indicating the type of the control parameter to be searched by the machine learning unit 19, and information indicating the range of values of the control parameter to be searched among the control parameters constituting the machining conditions. That is, the machine learning parameter includes information for defining a search range of the machining conditions related to the machine learning. The machine learning parameter includes information of a calculation formula for calculating an evaluation point of the processing condition. The evaluation points include learning data for machine learning.
The machine learning parameter determination unit 15 includes: a search parameter determination unit 16 that determines a control parameter to be set as a search target of the machine learning unit 19; a search range determination unit 17 that determines a range of values of control parameters to be searched; and an evaluation point calculation formula determination unit 18 that determines a calculation formula of the evaluation point.
The determination rule storage unit 14 holds information of a determination rule for determining a machine learning parameter based on a machining specification. The determination rule storage unit 14 stores information of a predetermined determination rule. The stored determination rule is updated by a determination rule updating unit 21 described later.
The search parameter determination unit 16 determines a control parameter to be a search target based on the machining specification, the determination rule, and the machining type indicated by the machine learning setting information. When the user specification information on the type of the control parameter is included in the machine learning setting information, the search parameter determination unit 16 further determines the control parameter to be the search target based on the user specification information.
The search range determining unit 17 determines a range of values of the control parameters to be searched based on the machining specification, the determination rule, and the machining type indicated by the machine learning setting information. The search range determining unit 17 adjusts the range of values of the control parameter based on the reference machining condition. When the user specification information on the range of the value of the control parameter is included in the machine learning setting information, the search range determination unit 17 further determines the range of the value of the control parameter based on the user specification information.
The evaluation point calculation formula determination unit 18 determines a calculation formula of an evaluation point based on the machining specification, the determination rule, and the machining type indicated by the machine learning setting information. Specifically, the evaluation point calculation formula determination unit 18 determines the value of the variable used in the calculation formula.
As described above, the machine learning parameter determination unit 15 determines the machine learning parameter based on the machining specification, the determination rule, the machine learning setting information, and the reference machining condition. The machine learning parameter determination unit 15 outputs machine learning parameters including information indicating the type of the parameter to be searched, information indicating the range of values of the parameter to be searched, and information of the determined calculation formula.
The machine learning unit 19 learns machining conditions for performing electric discharge machining by the electric discharge machine 28, and searches for machining conditions to be output to the electric discharge machine 28 based on the learning result. The reference machining conditions, machine learning parameters, and state data are input to the machine learning unit 19. The state data is data indicating a state of electric discharge machining in the electric discharge machine 28. The state data includes data of a machining depth, data of a machining speed, the number of times of discharge generation, pulse state data, and the like. The pulse state data is data indicating whether or not the state of the current pulse or the voltage pulse is acceptable. The state data is input from the electric discharge machine 28 to the machine learning unit 19.
The machine learning unit 19 includes a learning unit 22, a state observation unit 23, a machining condition adjustment unit 24, and an evaluation point calculation unit 25. The state observation unit 23 observes, as state variables, evaluation points, which are information indicating the type of the control parameter to be searched, information indicating the range of values of the control parameter to be searched, and information on the state of the electric discharge machining, among the machine learning parameters. The learning unit 22 learns an optimal combination of values of the control parameters according to the data set created based on the state variables. The learning unit 22 includes a return calculation unit 26 and a function update unit 27. The return calculation unit 26 and the function update unit 27 will be described later.
The machining condition adjustment unit 24 adjusts the machining conditions to be output to the electric discharge machine 28 based on the learning result obtained by the learning unit 22. The evaluation point calculating unit 25 calculates an evaluation point concerning the machining condition output to the electric discharge machine 28 based on the state data.
The machining condition searching device 10 outputs the machining conditions adjusted by the machining condition adjusting unit 24 to the electric discharge machine 28. The electric discharge machine 28 machines the workpiece in accordance with the machining conditions input from the machining condition search device 10.
The machining conditions adjusted by the machining condition adjusting unit 24 are input to the machining condition adjustment result storage unit 20. The machining condition adjustment result storage unit 20 holds the machining conditions adjusted by the machining condition adjustment unit 24.
In fig. 1, the machining condition search device 10 is shown as an external device to the electric discharge machine 28, but the machining condition search device 10 may be incorporated in the electric discharge machine 28. The machining condition search device 10 may be built in a numerical control device that controls the electric discharge machine 28.
The determination rule updating unit 21 reads out the machining conditions from the machining condition adjustment result storage unit 20. The determination rule updating unit 21 updates the determination rule of the machine learning parameter based on the adjustment result of the machining condition stored in the machining condition adjustment result storage unit 20.
Next, the processing performed by the processing condition search device 10 will be described. Fig. 2 is a flowchart showing a procedure of processing performed by the processing condition search device 10 according to embodiment 1.
The user inputs information of the machining specification to the machining specification input unit 11. Further, the user inputs machine learning setting information to the machine learning setting input unit 13. In step S1, the machining condition search device 10 receives machining specification information and machine learning setting information. The machine learning setting information includes information indicating a machining type. Alternatively, the machine learning setting information includes information indicating the type of machining and user-specified information.
In step S2, the machining condition search device 10 creates a reference machining condition corresponding to the machining specification in the machining condition creation unit 12. The machining condition creation unit 12 creates a reference machining condition by selecting a machining condition corresponding to the machining specification input unit 11 from among a plurality of machining conditions stored in advance.
In step S3, the machining condition searching apparatus 10 determines the machine learning parameter based on the machining specification, the determination rule of the machine learning parameter, the machine learning setting information, and the reference machining condition in the machine learning parameter determining unit 15. The machine learning parameter determination unit 15 determines machine learning parameters including information indicating the type of the parameter to be searched, information indicating the range of values of the parameter to be searched, and information of the calculation formula.
Here, a decision rule of the machine learning parameter will be described. Fig. 3 is a diagram showing an example of a determination rule used by the processing condition search device 10 according to embodiment 1 to determine the machine learning parameter. The decision rule is represented by an association of the process specification and the machine learning parameter. The determination rule storage unit 14 holds a database including information on the machining specifications and information on machine learning parameters associated with the machining specifications.
In fig. 3, 3 items of "processing type", "processing shape", and "important performance" are shown in the column of "processing specification". The information of the machining type is information included in the machine learning setting information, but in fig. 3, the machining type is handled as 1 item of the machining specification. In fig. 3, for example, in the rows where "machining type" is "rough machining", "machining shape" is "circular" and "important performance" is "speed", information of machine learning parameters corresponding to machining specifications of "rough machining", "circular" and "speed" is shown.
In the column of "search parameters" in fig. 3, each item of "rest time", "jump parameter 1", "jump parameter 2", "servo parameter 1" and "servo parameter 2" represents a control parameter. The "rest time" is a control parameter indicating the discharge and the interval time of the discharge. The "jump parameter 1" and the "jump parameter 2" are control parameters related to the jump motion of the electrode, respectively. The "servo parameter 1" and the "servo parameter 2" are parameters indicating the number of discharges per unit time or the interval between the electrode and the workpiece, respectively.
In fig. 3, the "Σ" in the column of the "search parameter" indicates a control parameter to be searched for. The "-" in the column of the "search parameter" indicates a control parameter that is not a target of search. In fig. 3, for example, 4 of the "machining specifications," the "machining type" is "rough machining," "the" machining shape "is" circular, "and" the "important performance" is "speed," and "the rest time," "the jump parameter 1," "the jump parameter 2," and "the servo parameter 1" are targets to be searched for.
In the column of the "search range" in fig. 3, each item of the "minimum value" and the "maximum value" of the values in the range where the values are searched and the "scale" of the values in the range where the values are searched are shown for each control parameter in the column of the "search parameter". In fig. 3, for example, the "minimum value" 3, "maximum value" 7, and "scale" 1 in the column of the "rest time" are the range of 3 to 7 for the search range of the value of the control parameter "rest time", and the search is shown on the scale of 1 in this range. In fig. 3, the columns of "jump parameter 2", "servo parameter 1", and "servo parameter 2" in the columns of "search range" are omitted, but data concerning the values to be searched are set for in the columns of "jump parameter 2", "servo parameter 1", and "servo parameter 2" in the same manner as in the cases of "rest time" and "jump parameter 1" shown in fig. 3. The "-" in the column of the "search range" indicates that no data on the value to be searched is set.
The evaluation point calculating unit 25 obtains data necessary for evaluating the machining conditions from the state data. For example, the evaluation point calculating unit 25 obtains the machining speed, the machining stability, and the pulse state data from the state data. The evaluation point calculation unit 25 calculates an evaluation point by substituting the machining speed, the machining stability, and the pulse state data into an evaluation formula. In the column of the "evaluation point calculation variable" in fig. 3, each item of the "machining speed", "machining stability", and "pulse state" indicates the value of the variable used in the calculation formula of the evaluation point. That is, the variables for each item represent weights in the evaluation of the processing conditions.
For example, since the processing speed is emphasized with respect to the processing conditions corresponding to the processing specifications in which the "processing type" is "rough processing", "processing shape" is "circular", and "important performance" is "speed", a high value of "0.8" is set as a variable of the "processing speed". In addition, the process stability can be ignored in the process specification, and "0" is set as a variable of "process stability". Further, from the viewpoint of the processing speed, in view of the fact that even if the evaluation is high, if the failure rate of the current pulse or the voltage pulse is high, the processing speed is lowered, and the variable of the "pulse state" is set to "0.2" for taking the pulse state into consideration when evaluating the processing condition.
In fig. 3, the items of "machining specification" in the determination rule are 3 items of "machining type", "machining shape" and "important performance", but the items of "machining specification" are not limited to these items. The item of the "processing specification" can be arbitrarily set.
In fig. 3, in the "search parameter" of the determination rule, 5 control parameters, i.e., whether or not the control parameter to be set as the search target is set to "rest time", "jump parameter 1", "jump parameter 2", "servo parameter 1", and "servo parameter 2", but the present invention is not limited thereto. Whether or not the control parameter to be the object of search is set in the "search parameter" can be arbitrarily set.
In fig. 3, the items that can set the "evaluation point calculation variable" are 3 items of the "machining speed", "machining stability", and "pulse state" in the determination rule, but the items of the "evaluation point calculation variable" are not limited to these items. The item of the "evaluation point calculation variable" can be arbitrarily set.
The machine learning parameter determination unit 15 reads out, from the database of the determination rule storage unit 14, information of "search parameters", "search range", and "evaluation point calculation variables" related to the machining type indicated by the input machine learning setting information, the machining shape indicated by the input machining specification information, and the important performance.
The search parameter determination unit 16 obtains information of "search parameters". When the user specification information on the type of the control parameter is included in the machine learning setting information, the search parameter determination unit 16 adjusts the type of the control parameter indicated by the "search parameter" based on the user specification information. Thereby, the search parameter determination unit 16 determines the control parameter to be searched for.
The search range determination unit 17 obtains information of the "search range". The search range determination unit 17 adjusts the range of values indicated by the "search range" based on the reference processing conditions. For example, the search range determining unit 17 adjusts the value of the control parameter so that the rest time is shortened and the time for the jump operation is shortened as compared with the reference machining condition. When the user specification information on the range of the value of the control parameter is included in the machine learning setting information, the search range determination unit 17 adjusts the range of the value indicated by the "search range" based on the user specification information. Thus, the search range determining unit 17 determines a range of values to be searched for each control parameter.
The evaluation point calculation formula determination unit 18 obtains information of "evaluation point calculation variables". Thus, the evaluation point calculation formula determination unit 18 determines the variables used in the calculation formula of the evaluation point. That is, the evaluation point calculation formula determination unit 18 determines the calculation formula of the evaluation point. The machine learning parameters including information indicating the type of the parameter to be searched, information indicating the range of values of the parameter to be searched, and information of the determined calculation formula are input from the machine learning parameter determination unit 15 to the machine learning unit 19.
The machining condition search device 10 inputs the machining specification to the machining specification input unit 11, and inputs the machine learning setting information including the information of the machining type to the machine learning setting input unit 13, thereby determining the machine learning parameters corresponding to the inputted machining specification and machining type. The machining condition search device 10 can determine machine learning parameters corresponding to an arbitrary machining specification and machining type specified by a user. The processing condition search device 10 can determine the machine learning parameter based on the specification of the user regarding the type of the control parameter and the value range of the control parameter by inputting the machine learning setting information including the user specification information to the machine learning setting input unit 13.
After the machine learning parameters are input to the machine learning unit 19, the machining condition search device 10 starts outputting the machining conditions from the machining condition adjustment unit 24 to the electric discharge machine 28 in step S4 shown in fig. 2. The electric discharge machine 28 machines the workpiece in accordance with the machining conditions input from the machining condition search device 10.
The reference machining conditions are input to the machining condition adjustment unit 24. When the electric discharge machine 28 starts machining, the machining condition adjustment unit 24 outputs the reference machining condition as a machining condition. The electric discharge machine 28 starts machining in accordance with the reference machining conditions.
After starting the machining by the electric discharge machine 28, the machining condition searching device 10 performs machine learning by the learning unit 22 in step S5. The machining condition search device 10 adjusts the machining conditions based on the machine learning result in the machining condition adjustment unit 24.
When the electric discharge machine 28 performs machining, state data from the electric discharge machine 28 is input to the machine learning unit 19. The evaluation point calculating unit 25 obtains machining speed, machining stability, and pulse state data from the state data. For example, the evaluation point calculating unit 25 estimates the machining speed from the displacement amount of the machining depth. The evaluation point calculating unit 25 estimates the machining depth based on the amount of movement of the axis for moving the electrode and the data of the occurrence of the discharge. The discharge occurrence state is, for example, the number of discharge occurrences. The evaluation point calculation unit 25 obtains pulse state data based on the result of determination of whether or not discharge is acceptable. The pulse state data is, for example, a value indicating a proportion of defective pulses of the current pulse or the voltage pulse. The machining stability indicates fluctuation in the number of times of discharge occurrence in units of a certain period, for example, in units of several ms. The evaluation point calculation unit 25 obtains the machining stability based on the number of discharge occurrences in units of a fixed period. The greater the fluctuation in the number of times of discharge occurrence, the lower the processing stability. For example, when machining is unstable due to an instantaneous increase in the number of times of discharge occurrence, it is estimated that there is a problem in control of machining or there is an abnormality in a state around the discharge position. The more uniform the number of discharge occurrences per a certain period, the higher the processing stability.
The evaluation point calculation unit 25 obtains information of a calculation formula input from the machine learning parameter determination unit 15 to the machine learning unit 19. The evaluation point calculation unit 25 calculates an evaluation point by substituting the machining speed, the machining stability, and the pulse state data into an evaluation formula.
The evaluation point calculated by the evaluation point calculating unit 25 and the machining condition to be evaluated are input to the state observing unit 23. Information indicating the type of the parameter to be searched among the machine learning parameters and information indicating the range of values of the parameter to be searched are input to the state observation unit 23. The state observation unit 23 observes, as state variables, evaluation points, processing conditions to be evaluated, information indicating the type of the parameter to be searched, and information indicating the range of values of the parameter to be searched.
The learning unit 22 learns an optimal combination of values of the control parameters according to the data set created based on the state variables. The learning unit 22 generates a learning model for searching for an optimal combination of values of the control parameters among the types of the parameters to be searched and the range of values of the parameters to be searched. The details of the machine learning performed by the learning unit 22 will be described later.
The learning model generated by the learning unit 22 is input to the machining condition adjustment unit 24. The machining condition adjustment unit 24 estimates a combination of values of the control parameters to be searched based on the learning model, and adjusts the machining condition. The machining condition searching device 10 outputs the machining conditions adjusted by the machining condition adjusting unit 24 to the electric discharge machine 28.
As described above, the machine learning unit 19 learns the machining conditions, and searches for the machining conditions output to the electric discharge machine 28 based on the learning result. For example, when the electric discharge machine 28 performs rough machining in accordance with the machining conditions input from the machining condition search device 10, the machine learning unit 19 always searches for the machining conditions so that the machining conditions become optimal until the electric discharge machine 28 finishes rough machining.
In step S6, the machining condition search device 10 stores the adjustment result of the machining condition obtained by the machining condition adjustment unit 24 in the machining condition adjustment result storage unit 20. The machining condition adjustment result storage unit 20 holds all the machining conditions adjusted by the machining condition adjustment unit 24.
In step S7, the machining condition searching apparatus 10 updates the determination rule of the machine learning parameter in the determination rule updating unit 21. The determination rule updating unit 21 reads out the machining conditions from the machining condition adjustment result storage unit 20, and updates the determination rule based on the adjustment result of the machining conditions. The processing condition search device 10 can search for the processing conditions more efficiently in the subsequent processing identical or similar to the processing for searching for the processing conditions by updating the determination rule. As described above, the processing condition search device 10 ends the processing according to the sequence shown in fig. 2.
Fig. 4 is a diagram showing an example of the search range of the processing conditions in the processing condition search device 10 according to embodiment 1. Fig. 4 shows examples of "minimum value", "maximum value" and "scale" concerning each control parameter of "rest time", "jump parameter 1" and "jump parameter 2". The left part of fig. 4 shows the range of values of the control parameters before the search range is defined in embodiment 1. The search range defined by embodiment 1 is shown in the right part of fig. 4.
Before the search range is defined, each of the "rest time", "jump parameter 1", and "jump parameter 2" can take any one of 16 values. In this case, the number of combinations of the respective values of "rest time", "jump parameter 1", and "jump parameter 2" becomes 4096.
In contrast, in embodiment 1, the search range is defined, and the number of values that can be obtained in the "rest time" is reduced from 16 to 5. The number of values that each of the "jump parameter 1" and "jump parameter 2" can take is reduced from 16 to 6. Thus, the number of combinations of the respective values of "rest time", "jump parameter 1", and "jump parameter 2" is reduced from 4096 to 180.
The machining condition search device 10 can search for machining conditions by limiting the search range of the machining conditions, and can reduce the number of times of observation of the repetitive machining state, that is, the number of tests, while adjusting the machining conditions. The machining condition search device 10 can search for the optimal machining condition accurately and efficiently.
When the machining specification identical to the machining specification input at the time of the previous machining is input, the learning unit 22 can update the learning model generated at the time of the previous machining by learning. Thus, the machining condition search device 10 can obtain a learning model with higher accuracy, and can search for machining conditions with higher accuracy.
Next, details of the machine learning performed by the learning unit 22 will be described. The learning unit 22 creates a data set in which the state variables input from the state observation unit 23 are integrated. The learning unit 22 learns the optimal combination of the values of the control parameters in the types of the parameters to be searched and the ranges of the values of the parameters to be searched, according to the data set. The learning algorithm used by the learning unit 22 may be any algorithm. As an example, a case will be described in which reinforcement learning (Reinforcement Learning) is applied to a learning algorithm used by the learning unit 22.
In reinforcement learning, an agent in an environment, i.e., a moving subject, observes a current state and determines actions to be taken. The agent reports from the environment by selecting actions, and learns the countermeasures that report most through a series of actions. As typical methods of reinforcement learning, Q learning (Q-learning), TD learning (TD-learning), and the like are known. For example, in the case of Q learning, an action value table, which is a general update formula of the action value function Q (s, a), is represented by the following formula (1). The action cost function Q (s, a) represents the value of an action, i.e., the action cost Q, of selecting the action "a" based on the environment "s".
[ 1 ]
Q(s t ，a t )←Q(s t ，a t )+α(r t+1 +γmax a Q(s t+1 ，a t )-Q(s t ，a t ))…(1)
In the above formula (1), "s t+1 "indicates the state of the environment at time t, a t The action at time t is shown. By action a t The environment becomes s t+1 。r t+1 The return by the change in the environment is represented, γ represents the discount rate, and α represents the learning coefficient. When Q learning is applied, the combination of values of the control parameters, which are the processing conditions, becomes the action "a t ”。
The update expression represented by the above expression (1) is to increase the action value Q if the action value of the best action "a" at the time "t+1" is larger than the action value Q of the action "a" executed at the time "t", and to decrease the action value Q in the opposite case. In other words, the action cost function Q (s, a) is updated so that the action cost Q of the action a at the time t approaches the best action cost at the time t+1. Thus, the best action value in an environment is propagated in turn to the action value in its previous environment.
The return calculation unit 26 calculates a return based on the state variable. The function updating unit 27 updates the function for determining the processing conditions, which is a combination of values of the control parameters, in accordance with the return calculated by the return calculating unit 26.
The return calculation unit 26 calculates the return "r" based on the evaluation point, which is the evaluation result of the state of the electric discharge machining. For example, if the evaluation point increases as a result of changing the value of the control parameter, the return calculation unit 26 increases the return "r". The return calculation unit 26 increases the return "r" by giving a return value of "1". The value of the return is not limited to "1". As a result of changing the value of the control parameter, the report calculation unit 26 decreases the report "r" when the evaluation point decreases. The return calculation unit 26 assigns a return value of "-1", thereby reducing the return "r". The value of the return is not limited to "-1".
The function updating unit 27 updates the function for determining the processing conditions in accordance with the return calculated by the return calculating unit 26. The updating of the function can be performed in accordance with the data set, for example by updating an action value table. An action value table is a data set stored in the form of a table that associates arbitrary actions with their action values. For example, in the case of Q learning, an action cost function Q(s) expressed by the above formula (1) t ，a t ) As a function for calculating a combination of values of the control parameters.
Fig. 5 is a flowchart showing a learning procedure performed by the learning unit 22 of the processing condition search device 10 according to embodiment 1. Referring to the flowchart of fig. 5, a reinforcement learning method of updating the action cost function Q (s, a) will be described.
In step S11, the learning unit 22 acquires a data set created based on the state variable. In step S12, the learning unit 22 calculates the return based on the evaluation point. In step S13, the learning unit 22 updates the action cost function Q (S, a) based on the return. In step S14, the learning unit 22 determines whether or not the action cost function Q (S, a) converges. The learning unit 22 determines that the action cost function Q (S, a) has converged based on the fact that the action cost function Q (S, a) is not updated in step S13.
When it is determined that the action cost function Q (S, a) does not converge (step S14, no), the learning unit 22 returns the sequence to step S11. When determining that the action cost function Q (S, a) has converged (Yes in step S14), the learning unit 22 ends the learning of the sequence shown in fig. 5. The learning unit 22 may return the sequence from step S13 to step S11 to continue learning without performing the determination in step S14.
The learning unit 22 outputs the generated learning model, which is the action cost function Q (s, a), to the machining condition adjustment unit 24. The machining condition adjustment unit 24 estimates a combination of values of the control parameters, which are objects to be searched, based on the learning model, and adjusts the machining condition.
In embodiment 1, the case where reinforcement learning is applied to the learning algorithm used by the learning unit 22 has been described, but learning other than reinforcement learning may be applied to the learning algorithm. The Learning unit 22 may perform machine Learning using a well-known Learning algorithm other than reinforcement Learning, for example, a Learning algorithm such as Deep Learning (Deep Learning), neural network, genetic programming, inductive logic programming, or support vector machine.
The learning unit 22 is not limited to the processing condition search device 10. The learning unit 22 may be implemented by an external device of the processing condition search device 10. In this case, the device functioning as the learning unit 22 may be a device that can be connected to the processing condition searching device 10 via a network. The device functioning as the learning unit 22 may be a device existing on a cloud server.
The learning unit 22 may learn the machining conditions according to the data sets created by the plurality of machining condition search devices 10. The learning unit 22 may create a data set based on data obtained from a plurality of processing condition search apparatuses 10 used at the same site, or may create a data set based on data obtained from a plurality of processing condition search apparatuses 10 used at different sites. The data set may be collected from a plurality of process condition discovery devices 10 operating independently of one another at a plurality of sites. After the collection of the data sets from the plurality of processing condition searching apparatuses 10 is started, a new processing condition searching apparatus 10 may be added to the object to which the data sets are collected. In addition, after the collection of the data sets from the plurality of processing condition searching apparatuses 10 is started, a part of the plurality of processing condition searching apparatuses 10 may be excluded from the objects to which the data sets are collected.
The learning unit 22 that learns the 1 processing condition searching device 10 may learn the processing condition searching devices 10 other than the processing condition searching device 10. The learning unit 22 that performs learning related to the other processing condition search device 10 can update the learning model by relearning in the other processing condition search device 10.
Next, a hardware configuration of the processing condition search device 10 will be described. Fig. 6 is a diagram showing a configuration example of hardware of the processing condition search device 10 according to embodiment 1. The configuration example shown in fig. 6 is a configuration example in the case where the essential part of the processing condition search device 10 is realized by the processing circuit 30 having the processor 32 and the memory 33. The machining condition creation unit 12, the machine learning parameter determination unit 15, the machine learning unit 19, and the determination rule update unit 21 shown in fig. 1 are essential units of the machining condition search device 10.
The processor 32 is CPU (Central Processing Unit). The processor 32 may be an arithmetic device, a microprocessor, a microcomputer, or DSP (Digital Signal Processor). The memory 33 is, for example, RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (registered trademark) (Electrically Erasable Programmable Read Only Memory), or the like.
The memory 33 stores a machining condition search program that is a program for operating as a main part of the machining condition search device 10. The processor 32 reads and executes the machining condition search program, thereby realizing the essential part of the machining condition search device 10. The processing condition search program stored in the memory 33 may be provided in a state of being written in a storage medium such as CD (Compact Disc) -ROM or DVD (Digital Versatile Disc) -ROM, for example, or may be provided via a communication line.
Instead of the processor 32 and the memory 33, the essential part of the processing condition searching device 10 may be realized by a dedicated processing circuit. The dedicated processing circuit is a single circuit, a composite circuit, ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), or a circuit combining them. In addition, as for the essential part of the processing condition searching device 10, a part of the functions may be realized by the processor 32 and the memory 33, and the remaining functions may be realized by a dedicated processing circuit.
The input unit 31 is a circuit for receiving an input signal to the processing condition search device 10 from the outside. The state data from the electric discharge machine 28 is input to the input unit 31. The input unit 31 includes an input device for receiving an input operation. The machining specification input unit 11 and the machine learning setting input unit 13 are realized by input means.
The output unit 35 is a circuit that outputs the signal generated by the machining condition search device 10 to the outside of the machining condition search device 10. The output unit 35 outputs the machining conditions adjusted by the machining condition search device 10 to the electric discharge machine 28.
The storage device 34 stores various information. The storage device 34 is HDD (Hard Disk Drive) or SSD (Solid State Drive) or the like. The determination rule storage unit 14 and the machining condition adjustment result storage unit 20 are realized by a storage device 34.
According to embodiment 1, the machining condition search device 10 learns the machining conditions for electric discharge machining, and searches for the machining conditions to be output to the electric discharge machine 28 based on the learning result. The processing condition search device 10 determines a machine learning parameter indicating a control parameter to be searched and a range of values of the control parameter to be searched based on the processing specification. The machining condition search device 10 determines machine learning parameters, and searches for machining conditions based on learning of the determined machine learning parameters, thereby efficiently searching for machining conditions suitable for a machining state. As described above, the machining condition search device 10 has an effect of enabling efficient search of the machining conditions for electric discharge machining.
The configuration shown in the above embodiment shows an example of the content of the present invention. The structure of the embodiment can be combined with other known techniques. A part of the structure of the embodiment may be omitted or changed without departing from the scope of the present invention.
Description of the reference numerals
The machining condition searching device comprises a machining condition searching device 10, a machining specification input part 11, a machining condition creating part 12, a machine learning setting input part 13, a machine learning setting input part 14, a machine learning parameter determining part 15, a searching parameter determining part 16, a searching range determining part 17, an evaluation point calculating type determining part 18, a machine learning part 19, a machining condition adjusting result storing part 20, a determining rule updating part 21, a learning part 22, a state observing part 23, a machining condition adjusting part 24, a evaluation point calculating part 25, a reporting calculating part 26, a function updating part 27, a discharge machining machine 28, a processing circuit 30, an input part 31, a processor 32, a memory 33, a memory 34 and an output part 35.
Claims (11)
1. A processing condition searching device is characterized by comprising:
a machine learning unit that learns machining conditions for electric discharge machining performed by an electric discharge machine, and searches for machining conditions output to the electric discharge machine based on a learning result; and
And a machine learning parameter determination unit that determines a machine learning parameter indicating a range of values of a control parameter set as a target of search by the machine learning unit and a control parameter set as a target of search among control parameters constituting the machining condition, based on a machining specification specified in relation to the electric discharge machining.
2. The processing condition searching device according to claim 1, wherein,
the machine learning unit includes:
a state observation unit configured to observe, as state variables, information indicating a type of a control parameter to be searched, information indicating a range of values of the control parameter to be searched, and information on a state of the electric discharge machining among the machine learning parameters;
a learning unit that learns an optimal combination of values of control parameters according to a data set created based on the state variables; and
and a machining condition adjustment unit that adjusts a machining condition to be output to the electric discharge machine based on the learning result obtained by the learning unit.
3. The processing condition searching device according to claim 2, wherein,
The machine learning section includes an evaluation point calculation section that calculates an evaluation point concerning a machining condition output to the electric discharge machine based on state data indicating a state of the electric discharge machining,
the state observation unit observes the evaluation point, which is information on the state of the electric discharge machining, as the state variable.
4. The processing condition searching device according to claim 3, wherein,
the machine learning parameter determination unit determines the machine learning parameter including information of a calculation formula for calculating the evaluation point based on the machining specification.
5. The processing condition searching device according to any one of claims 2 to 4,
the processing apparatus includes a determination rule storage unit that holds a determination rule for determining the machine learning parameter based on the processing specification.
6. The processing condition searching device according to claim 5, wherein,
the processing condition adjustment device is provided with a determination rule update unit which updates the determination rule based on the adjustment result of the processing condition obtained by the processing condition adjustment unit.
7. The processing condition searching device according to any one of claims 1 to 6, wherein,
the processing specification input unit is provided with a processing specification input unit which inputs information of the processing specification.
8. The processing condition searching device according to claim 7, wherein,
comprises a machining condition creation unit for creating a reference machining condition, which is a machining condition composed of initial values of control parameters, based on the machining specification,
the machine learning parameter determination unit determines the machine learning parameter based on the machining specification and the reference machining condition.
9. The processing condition searching device according to claim 7, wherein,
comprises a machine learning setting input unit for inputting machine learning setting information indicating a machining type to be searched for as a machining condition,
the machine learning parameter determination unit determines the machine learning parameter based on the machining specification and the machine learning setting information.
10. The processing condition searching device according to claim 9, wherein,
the machine learning setting information includes information for specifying, by a user of the processing condition search device, a type of a control parameter to be searched or a range of values of the control parameter to be searched.
11. A machining condition searching method for searching for machining conditions for electric discharge machining by an electric discharge machine,
the processing condition searching method is characterized by comprising the following steps:
determining a machine learning parameter indicating a control parameter to be searched among control parameters constituting the machining condition and a range of values of the control parameter to be searched, based on a machining specification specified with respect to the electric discharge machining method;
learning a machining condition based on the machine learning parameter; and
and adjusting the machining conditions output to the electric discharge machine based on the learning result.
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2021/023335 WO2022269664A1 (en) | 2021-06-21 | 2021-06-21 | Machining condition searching device and machining condition searching method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116490314A true CN116490314A (en) | 2023-07-25 |
Family
ID=84544316
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202180079088.9A Pending CN116490314A (en) | 2021-06-21 | 2021-06-21 | Processing condition searching device and processing condition searching method |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP7204952B1 (en) |
| CN (1) | CN116490314A (en) |
| WO (1) | WO2022269664A1 (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111812633A (en) * | 2019-11-27 | 2020-10-23 | 谷歌有限责任公司 | Detecting reference frame changes in smart device based radar systems |
| CN111954582A (en) * | 2018-04-17 | 2020-11-17 | 三菱电机株式会社 | Machine learning device, electric discharge machine, and machine learning method |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6140228B2 (en) * | 2015-08-27 | 2017-05-31 | ファナック株式会社 | Wire electrical discharge machine for machining while adjusting machining conditions |
| EP3446820A1 (en) * | 2017-08-22 | 2019-02-27 | Agie Charmilles SA | Wire electrical discharge machining method |
| JP6860540B2 (en) * | 2018-10-25 | 2021-04-14 | ファナック株式会社 | Output device, control device, and learning parameter output method |
| JP6978452B2 (en) * | 2019-02-12 | 2021-12-08 | ファナック株式会社 | How to set the machine learning device, control device, and machine learning search range |
| CN113993646B (en) * | 2019-06-21 | 2023-01-10 | 三菱电机株式会社 | Machining condition search device and wire electric discharge machine |
| DE112020006911T5 (en) * | 2020-05-20 | 2023-01-12 | Mitsubishi Electric Corporation | Data generation device, machine learning system and machining state estimation device |
-
2021
- 2021-06-21 CN CN202180079088.9A patent/CN116490314A/en active Pending
- 2021-06-21 JP JP2021565826A patent/JP7204952B1/en active Active
- 2021-06-21 WO PCT/JP2021/023335 patent/WO2022269664A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111954582A (en) * | 2018-04-17 | 2020-11-17 | 三菱电机株式会社 | Machine learning device, electric discharge machine, and machine learning method |
| CN111812633A (en) * | 2019-11-27 | 2020-10-23 | 谷歌有限责任公司 | Detecting reference frame changes in smart device based radar systems |
Also Published As
| Publication number | Publication date |
|---|---|
| JPWO2022269664A1 (en) | 2022-12-29 |
| WO2022269664A1 (en) | 2022-12-29 |
| JP7204952B1 (en) | 2023-01-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6140228B2 (en) | Wire electrical discharge machine for machining while adjusting machining conditions | |
| US10921774B2 (en) | Machine learning devices and methods for optimizing the speed and accuracy of thread mill, inner diameter, outer shape, and surface machine tools | |
| US10121107B2 (en) | Machine learning device and method for optimizing frequency of tool compensation of machine tool, and machine tool having the machine learning device | |
| CN108241342B (en) | Numerical controller and machine learning device | |
| JP6063013B1 (en) | Numerical control device with machining condition adjustment function to suppress chatter or tool wear / breakage | |
| CN109581962B (en) | Numerical control system | |
| JP6903240B1 (en) | Motor control device and motor control method | |
| JP6348098B2 (en) | Wire EDM machine simulation device with core welding position determination function using machine learning | |
| US11554448B2 (en) | Machining condition adjustment apparatus and machine learning device | |
| JP2017030067A (en) | Control device-added machining apparatus with machining time measuring function and on-machine measuring function | |
| US11640557B2 (en) | Machine learning device, numerical control system, and machine learning method | |
| CN110355462B (en) | Machining condition adjustment device and machine learning device | |
| KR102224970B1 (en) | Controller and machine learning device | |
| CN111954582B (en) | Machine learning device | |
| CN110376964A (en) | Machine learning device, control device and machine learning method | |
| US20220269241A1 (en) | Numerical control device, machine learning device, and numerical control method | |
| JP2020123191A (en) | Numeric control system | |
| CN116490314A (en) | Processing condition searching device and processing condition searching method | |
| CN113993646B (en) | Machining condition search device and wire electric discharge machine | |
| JPWO2021111530A1 (en) | Control device, electric discharge machine, and machine learning device | |
| CN118019614A (en) | Numerical control device, machining system, numerical control method and machining method | |
| JPWO2016194072A1 (en) | Method for estimating machining time of wire electric discharge machine and control device for wire electric discharge machine | |
| JP7313585B1 (en) | Drive condition determination device and drive condition determination method | |
| WO2022219670A1 (en) | Machining condition setting device, machining condition setting method, and electrical discharge machining apparatus | |
| WO2024122066A1 (en) | Parameter adjustment device and computer-readable recording medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |