JP5734086B2 - Machining condition search device - Google Patents

Description

  The present invention relates to a machining condition search device for searching for an appropriate machining condition.

A processing machine used in industrial applications, for example, performs a desired process by gradually changing the shape of a metal workpiece made of tool steel by applying a physical, electrical, or chemical action. A workpiece having the final shape is produced.
The workpieces to be processed include not only metals but also plastics, wood, organic matter and glass. Also, the initial shape of the workpiece is various such as a rod shape, a plate shape, a lump shape, and a cylindrical shape.

Generally, a processing machine can change the state of a processing process by changing a control parameter called a processing condition. Therefore, selecting an appropriate processing condition determines the success or failure of processing.
However, it is actually difficult to select an appropriate machining condition that satisfies the required machining specifications for the workpiece.

For example, there is laser processing that obtains a desired shape by using laser beam energy or oxidation combustion reaction.
In this laser processing, select the processing conditions that satisfy the required surface roughness and shape accuracy, and maintain an appropriate processing state without being affected by aging of control parameters or variations in workpiece components. It is demanded.
This processing condition is composed of a plurality of control parameters, and each control parameter is configured to select one of values changing in several stages.
For example, when there are nine types of control parameters and each control parameter can select a value in nine stages, one is selected from 387420489 (= 9 ⁹ ).

In order to select the appropriate machining conditions that satisfy the required machining specifications, it is assumed that the machining conditions are comprehensively designed in advance, but it is not easy to design the machining conditions comprehensively. Absent.
The reason is that when acquiring the processing characteristics of the processing machine in a prior experiment, the number of combinations of control parameters is very large, and it is costly and time consuming to perform an experiment under all processing conditions.

In order to design the machining conditions, it is necessary to construct a machining characteristic model that indicates the relationship between the machining conditions of the workpiece and the machining result. However, it is difficult and difficult to construct the machining characteristic model.
The reason is that it is difficult to strictly determine whether the processing result is good or bad. For example, factors that determine the quality of processing in laser processing include not only whether or not cutting is possible, but also the roughness of the processed surface and the presence or absence of dross, but these evaluations depend on the judgment of the operator's appearance. Therefore, quantitative evaluation cannot be performed and evaluation may be wrong. In addition, there is a method of measuring a light emission waveform, light emission intensity, and sound wave at the time of processing and estimating a processing result based on the measured value, but none of them is an index that completely captures the processing phenomenon, and the evaluation may be wrong.
For this reason, a machining characteristic model is generated by evaluation of an incorrect machining result, and an undesirable machining condition may be designed.

Therefore, it is desired to develop a method for generating a highly accurate machining characteristic model and a method for selecting an appropriate machining condition that satisfies the required machining specifications.
In Patent Document 1 below, a machining characteristic model is heuristically constructed based on the knowledge obtained from the results of machining experiments, and required specifications relating to machining to be performed are input to the machining characteristic model. Thus, a method of outputting machining conditions that satisfy the required specifications is disclosed.
Patent Document 2 below shows a method for making a machining characteristic empirical formula nonlinear by fuzzy inference.

Patent Document 3 below discloses a method of correcting a sequential machining characteristic model based on the result of experimental machining.
In Patent Document 4 below, instead of preparing a machining characteristic model in advance, a machining characteristic model is sequentially estimated by repeating experimental machining, and experimental machining conditions are newly generated from the estimated machining characteristic model. A method for generating a highly accurate machining characteristic model with a small number of experiments is shown.

JP 2009-282909 A Japanese Patent No. 2880964 Japanese Patent No. 406852 JP 2010-42499 A

  Since the conventional machining condition search apparatus is configured as described above, if the method disclosed in Patent Document 4 is applied, a machining characteristic model with high accuracy can be generated with a small number of experiments. However, when there is an error in the quality evaluation of the machining result, there is a problem that the influence of the error cannot be eliminated and an appropriate machining condition cannot be generated.

  The present invention has been made to solve the above-described problems, and it is possible to search for the best machining conditions with a small number of experiments, and even if there is an error in the quality evaluation of the machining result, the influence of the error An object of the present invention is to obtain a machining condition search device that can eliminate the above and generate an appropriate machining condition.

The machining condition search device according to the present invention generates an experimental machining condition that generates an experimental machining condition that is a combination of a plurality of control parameters, using a machining characteristic model that indicates a relationship between a machining condition of a workpiece to be machined and a machining result. And machining results of experimental machining are collected from a processing machine that experimentally processes the workpiece according to the experimental machining conditions generated by the experimental machining condition generation means, and a set of the machining results and the above experimental machining conditions is accumulated as experimental machining data A first machining characteristic model for generating a new machining characteristic model using the machining result collecting means to perform and the experimental machining data accumulated by the machining result collecting means, and outputting the machining characteristic model to the experimental machining condition generation means. For each experimental machining data accumulated by the generating means and machining result collecting means, the machining quality included in the machining results in the experimental machining data Evaluation to failure from good one by one, or while changing from bad to good, and the second processing characteristics model generation means for generating a new processing properties model processing quality evaluation after the change is reflected, a second Machining characteristic model synthesizing means for synthesizing a new machining characteristic model generated by the machining characteristic model generating means, and the optimum machining condition generating means optimizes machining from the machining characteristic model synthesized by the machining characteristic model synthesizing means. A condition is generated.

According to the present invention, an experimental machining condition generating means for generating an experimental machining condition that is a combination of a plurality of control parameters using a machining characteristic model indicating a relationship between a machining condition of a workpiece to be machined and a machining result, and an experiment Machining result collection that collects machining results of experimental machining from a machine that performs experimental machining of workpieces according to the experimental machining conditions generated by the machining condition generation means, and accumulates a set of the machining results and the above experimental machining conditions as experimental machining data And a first machining characteristic model generation unit that generates a new machining characteristic model using the experimental machining data accumulated by the machining result collection unit and outputs the machining characteristic model to the experimental machining condition generation unit, each accumulated experimental processing data by the processing result collecting means, one by one machining quality evaluation included in the processing result of the within experimental processed data From good to bad, or while changing from bad to good, and the second processing characteristics model generation means for generating a new processing properties model processing quality evaluation after the change is reflected, a second processing characteristics model generation Machining characteristic model synthesizing means for synthesizing a new machining characteristic model generated by the means, and the optimum machining condition generating means generates an optimum machining condition from the machining characteristic model synthesized by the machining characteristic model synthesizing means. As a result, the best machining conditions can be searched with a small number of experiments, and even if there is an error in the quality evaluation of machining results, the influence of the error can be eliminated and appropriate machining conditions can be generated. There is an effect that can.

It is a block diagram which shows the processing condition search apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the processing condition search apparatus by Embodiment 1 of this invention. It is explanatory drawing which shows an example of the experimental process data accumulate | stored by the actual process result storage part. It is explanatory drawing which has shown notionally the structure of the process characteristic model. It is explanatory drawing which shows an example of the process characteristic model with respect to two parameters a and b. It is explanatory drawing which shows the calculation method of an expected value. It is explanatory drawing which shows the change of the process characteristic model accompanying the partial change of process quality evaluation. It is explanatory drawing which shows an example of an operator interface screen. It is explanatory drawing which shows the selection method of the process result which changes process quality evaluation.

Embodiment 1 FIG.
1 is a block diagram showing a machining condition searching apparatus according to Embodiment 1 of the present invention.
The processing condition search apparatus in FIG. 1 includes an experimental processing condition generation unit 11, a processing result collection unit 12, a processing characteristic model change unit 13, and an optimum processing condition generation unit 14. The experimental processing condition generation unit 11, the processing result Each of the collecting unit 12, the machining characteristic model changing unit 13, and the optimum machining condition generating unit 14 is configured by dedicated hardware (for example, a semiconductor integrated circuit on which a CPU is mounted or a one-chip microcomputer). Assumed.
However, the machining condition search device may be configured by a computer. When the machining condition search device is configured by a computer, an experimental machining condition generation unit 11, a machining result collection unit 12, and a machining characteristic model change unit 13 are provided. And the program which describes the processing content of the optimal process condition production | generation part 14 may be stored in the memory of a computer, and CPU of the said computer may be made to run the program stored in the said memory.
FIG. 2 is a flowchart showing the processing contents of the machining condition search device according to Embodiment 1 of the present invention.

In FIG. 1, the processing machine 1 experimentally processes a workpiece to be processed according to the experimental processing conditions generated by the experimental processing condition generation unit 11, and processing results of the experimental processing (phenomenon and information obtained during the experimental processing, This is an industrial device that outputs phenomena and information obtained after the end of experimental processing to the processing result collection unit 12.
For example, the processing machine 1 removes, adds, or deforms a desired shape from a lump of metal to finally obtain a desired shape to produce a product (workpiece). Such a processing machine 1 As such, a cutting machine, a laser machine, an electric discharge machine, an electron beam machine, a plasma machine, an electrolytic machine, a forging machine, a rolling machine, a welding machine, a surface treatment machine, and the like are applicable.
In the first embodiment, description will be made assuming that the processing machine 1 is a laser processing machine.

The experimental machining condition generation unit 11 includes an experimental machining condition calculation unit 21, an experimental machining condition storage unit 22, a machining characteristic model generation unit 25, and a machining characteristic model storage unit 26. Material, workpiece shape, type of tool used for machining, dimensions, etc.) and machining characteristics model that shows the relationship between workpiece machining conditions and machining results, and generates experimental machining conditions that are combinations of multiple control parameters. And the process which gives the experimental processing conditions to the processing machine 1 is implemented.
The machining result collection unit 12 includes a machining result determination unit 23 and an actual machining result storage unit 24, collects machining results of experimental machining from the processing machine 1, and outputs the machining results and the experimental machining condition generation unit 11. The process of accumulating the set of experimental machining conditions as experimental machining data is performed. The processing result collection unit 12 constitutes a processing result collection unit.

  The machining characteristic model change unit 13 includes a machining result evaluation partial change unit 27, a machining characteristic model generation unit 28, a machining characteristic model storage unit 29, a machining characteristic model synthesis unit 30, and a synthetic machining characteristic model storage unit 31. For each experimental machining data accumulated by the machining result collection unit 12, the machining quality evaluation after the change is reflected while changing the machining quality evaluation included in the machining results in the experimental machining data one by one. A process for generating a new machining characteristic model and synthesizing the new machining characteristic model is performed.

  The optimum machining condition generation unit 14 includes an optimum machining condition calculation unit 32 and an optimum machining condition storage unit 33, and performs processing for generating optimum machining conditions from the machining characteristic model synthesized by the machining characteristic model change unit 13. To do. The optimum machining condition generation unit 14 constitutes optimum machining condition generation means.

The experimental machining condition calculation unit 21 uses the machining characteristic model stored in the machining characteristic model storage unit 26 to generate an experimental machining condition that is a combination of a plurality of control parameters, and the experimental machining condition is stored in the experimental machining condition storage unit. The process which outputs to 22 is implemented.
Further, the experimental machining condition calculation unit 21 performs a process of determining whether or not to continue the experimental machining, and stops generating the experimental machining conditions when it is determined that the repetition of the experimental machining is to be stopped. As a criterion for determining whether or not to continue the experimental machining, for example, the number of experiments, time, the amount of change in the machining characteristic model, the absolute value of an expected value described later, and the like can be considered. The experimental machining condition calculation unit 21 constitutes experimental machining condition generation means.
The experimental machining condition storage unit 22 includes a storage device such as a RAM or a hard disk, and stores the experimental machining conditions output from the experimental machining condition calculation unit 21.

The machining result determination unit 23 performs a process of evaluating the quality of machining by digitizing the machining result output from the processing machine 1 using, for example, an automatic measuring device.
Specifically, the amplitude of the light emission waveform, the intensity of the light emission waveform, the sound wave, etc. at the time of machining are measured and digitized, and the binary evaluation of the numeric value is performed, for example, whether or not the cutting has been performed, whether the machining surface is rough. Then, the presence or absence of dross is judged.
The actual machining result storage unit 24 stores the experimental machining conditions used for the experimental machining, the required machining specifications for the workpiece, and the combination of the machining results as experimental machining data each time the machining machine 1 performs the experimental machining. It is.

Here, FIG. 3 is an explanatory diagram showing an example of experimental machining data accumulated by the actual machining result accumulation unit 24.
In the example of FIG. 3, one processing result is held as a set of a plurality of items (fields) such as time and place.
In addition, a set of experimentally processed data is represented by a table, and data of other tables are combined by an item ID number.
The experimental machining data accumulated by the actual machining result accumulation unit 24 is used when the format of the machining characteristic model generated by the machining characteristic model generation unit 25 and the parameters of the machining characteristic model format are determined.

The machining characteristic model generation unit 25 generates a new machining characteristic model using the experimental machining data accumulated by the actual machining result accumulation unit 24, and outputs a new machining characteristic model to the machining characteristic model storage unit 26. carry out. The machining characteristic model generation unit 25 constitutes first machining characteristic model generation means.
The machining characteristic model storage unit 26 includes a storage device such as a RAM or a hard disk, and stores a new machining characteristic model generated by the machining characteristic model generation unit 25.

The machining result evaluation partial changing unit 27 performs a process of changing the machining quality evaluation included in the machining result in the experimental machining data one by one for each experimental machining data accumulated by the actual machining result accumulation unit 24. To do.
Each time the machining result evaluation partial change unit 27 changes the machining quality evaluation, the machining characteristic model generation unit 28 generates a new machining characteristic model reflecting the changed machining quality evaluation, and a new machining characteristic model is created. Is output to the machining characteristic model storage unit 29.
The machining result evaluation partial change unit 27 and the machining characteristic model generation unit 28 constitute a second machining characteristic model generation unit.
The machining characteristic model storage unit 29 includes a storage device such as a RAM or a hard disk, and stores a new machining characteristic model generated by the machining characteristic model generation unit 28.

The machining characteristic model synthesis unit 30 synthesizes a new machining characteristic model accumulated by the machining characteristic model accumulation unit 29 and outputs a synthesized machining characteristic model, which is a machining characteristic model after synthesis, to the synthesized machining characteristic model storage unit 31. Perform the process. Note that the machining characteristic model synthesis unit 30 constitutes machining characteristic model synthesis means.
The composite processing characteristic model storage unit 31 is configured by a storage device such as a RAM or a hard disk, and stores the composite processing characteristic model output from the processing characteristic model combining unit 30.

The optimum machining condition calculation unit 32 calculates the optimum machining condition from the synthesized machining characteristic model stored in the synthesized machining characteristic model storage unit 31 and outputs the optimum machining condition to the optimum machining condition storage unit 33. .
The optimum machining condition storage unit 33 is composed of a storage device such as a RAM or a hard disk, for example, and stores the optimum machining conditions output from the optimum machining condition calculation unit 32.

Next, the operation will be described.
First, the experimental machining condition calculation unit 21 selects a predetermined number of experimental machining conditions as initial experimental machining conditions from a plurality of experimental machining conditions prepared in advance.
For example, in the case where experimental processing conditions are prepared in which seven types of control parameters that change to two levels are combined (total 7 to the seventh power (2 ⁷ = 128) experimental processing conditions are prepared. ), A predetermined number (for example, eight) of experimental machining conditions are selected as the initial experimental machining conditions from the experimental machining conditions.
The selection of the initial experimental machining conditions uses a so-called L8 orthogonal table, but is not limited to the L8 orthogonal table, but is an L4 orthogonal table, an L16 orthogonal table, an L32 orthogonal table, and an L64 orthogonal table. , L9 type orthogonal table, L27 type orthogonal table, L81 type orthogonal table, L243 type orthogonal table, L12 type orthogonal table, L18 type orthogonal table, L36 type orthogonal table, L72 type orthogonal table, and the like may be used.
Or multi-way layout, Latin square, Greco Latin square, Uniform design, Optimal design such as D-optimization, G-optimization, A-optimal, Compound design in response surface design, Box and Behnken design, or Totally random design May be used as appropriate.

For example, when the eight experimental processing conditions are selected as the initial experimental processing conditions, the experimental processing condition calculation unit 21 selects one initial experimental processing condition from the initial experimental processing conditions, and the experimental processing condition storage unit 22. Then, the initial experimental machining conditions are output to the processing machine 1 and the actual machining result storage unit 24 (step ST1 in FIG. 2).
When receiving the initial experimental machining condition from the experimental machining condition calculation unit 21, the processing machine 1 performs experimental machining on the workpiece according to the initial experimental machining condition, and outputs the machining result of the experimental machining to the machining result determination unit 23 (step). ST2).

When receiving the processing result from the processing machine 1, the processing result determination unit 23 quantifies the processing result using, for example, an automatic measuring device, and evaluates the quality of the processing (step ST3).
For example, by measuring the operation signal of the processing machine observed during processing (the amplitude of the light emission waveform at the time of processing, the intensity of the light emission waveform, the sound wave, etc.) and digitizing it, and performing binary evaluation of the numerical value, for example, Then, it is determined whether or not cutting is possible, the roughness of the processed surface, the presence or absence of dross, and the like.
As shown in FIG. 3, the actual machining result accumulating unit 24 accumulates the experimental machining conditions used for the experimental machining, the required machining specifications for the workpiece, and the combination of the machining results as experimental machining data (step ST4).

Thereafter, the experimental machining condition calculation unit 21 sequentially selects the remaining seven initial experimental machining conditions (step ST1), so that the processing machine 1 is set to the initial experimental machining conditions selected by the experimental machining condition calculation unit 21. Therefore, the workpiece is experimentally processed (step ST2).
Then, the machining result determination unit 23 evaluates the quality of machining (step ST3), and the actual machining result storage unit 24 accumulates experimental machining data (step ST4).
When the experimental machining based on the eight initial experimental machining conditions is completed (step ST5: YES), the process proceeds to step ST6.
Here, each time the processing machine 1 experimentally processes the workpiece, the actual processing result storage unit 24 stores the experimental processing data. However, after the experimental processing based on the eight types of initial experimental processing conditions is completed. 8 types of processing results may be collected and experimental processing data may be accumulated.

When the experimental machining based on the eight types of initial experimental machining conditions is completed, the machining characteristic model generation unit 25 generates a new machining characteristic model using the experimental machining data accumulated by the actual machining result accumulation unit 24, and creates a new machining characteristic model. The machining characteristic model is output to the machining characteristic model storage unit 26 (step ST6).
Here, FIG. 4 is an explanatory view conceptually showing the structure of the machining characteristic model.
In the example of FIG. 4, the machining characteristic model includes three functions that have machining conditions as arguments.
At the stage of searching for the processing conditions, it is possible to estimate the surface roughness of the processed surface, adhesion of dross, and whether or not processing is possible by using these functions. Each function can change the input / output relationship of the machining characteristic model according to separately defined operation parameters.
The operation parameter is determined by processing the experimental machining data accumulated by the actual machining result accumulation unit 24. For example, using a combination of function models and parameter determination algorithms such as multiple regression analysis of experimentally processed data, support vector machine and support vector regression, kernel regression, Bayesian point machine, Bayesian estimation, maximum likelihood estimation, response surface, neural network Done in

FIG. 5 is an explanatory diagram showing an example of a machining characteristic model for two parameters a and b.
In the example of FIG. 5, the points represented by ◯ and X are experimental machining points, and the line drawn with a solid line indicates that the machining result estimated by the machining characteristic model is good or defective. It shows the boundary.
The solid line represents the most likely estimation model (processing characteristic model) from the set of combinations of the parameters a and b obtained at the time and the determination result by experiment using the above-described method. Obtained by calculation. This estimation model is calculated and regenerated every time an experimental machining result is obtained.

When the machining characteristic model generation unit 25 outputs a new machining characteristic model to the machining characteristic model storage unit 26, the experimental machining condition calculation unit 21 generates the next experimental machining condition using the machining characteristic model (step ST7). ).
The following experimental machining conditions are the machining conditions expected to be the most effective in knowing the machining characteristics.
Hereinafter, the process for generating the experimental machining conditions will be specifically described.

FIG. 6 is an explanatory diagram showing a method for calculating an expected value.
A candidate point to be tested next is selected from a set (a, b) of combinations of machining conditions that have not been tested yet. The candidate points may be all the remaining points that have not been tested yet, or an estimation model. The candidate points may be narrowed down according to some criterion such as a certain distance from the boundary (solid line) obtained by the above.
By narrowing down the candidate points, the number of times the expected value is calculated is reduced, so that the calculation time can be shortened.

Next, for all of the selected candidate points, using the estimation model calculation method, a boundary (two-dot chain line in FIG. 6) drawn when it is determined that the processing result at the candidate points is good, A boundary (a chain line in FIG. 6) drawn when it is determined that the processing result at the candidate point is defective is calculated.
Next, the area A1 of the region surrounded by the solid line and the one-dot chain line is calculated, and the area A2 of the region surrounded by the solid line and the two-dot chain line is calculated.
And using area A1, A2, let A1 + A2 be an expected value.
Alternatively, an operation result of an expression such as (A1 + A2) / | A1-A2 + 1 |

At this time, a large A1 + A2 indicates that there is a high possibility that the estimated model is greatly changed, and a small | A1-A2 + 1 | indicates that a difference in the amount of change in the estimated model due to a difference in processing results of candidate points. It is small.
If | A1-A2 + 1 | is large, the amount of change in the estimated model increases, but if it is removed, the amount of change in the estimated model is small. If A1 + A2 is large, | A1-A2 + 1 | Since it is not good, it is effective to judge by looking at the calculation result of (A1 + A2) / | A1-A2 + 1 |.
When the expected value is calculated for all candidate points, the candidate point having the largest expected value is selected from the obtained expected values, and the candidate point is set as the next experimental processing condition, and the processing machine 1 and the actual value are selected. The result is output to the machining result storage unit 24.

  When the processing machine 1 receives the next experimental processing condition from the experimental processing condition calculation unit 21, the processing machine 1 experimentally processes the workpiece according to the experimental processing condition and outputs the processing result of the experimental processing to the processing result determination unit 23 (step). ST8).

When receiving the processing result from the processing machine 1, the processing result determination unit 23 quantifies the processing result using, for example, an automatic measuring device, and evaluates the quality of the processing (step ST9).
As shown in FIG. 3, the actual machining result accumulating unit 24 accumulates the experimental machining conditions used for the experimental machining, the required machining specifications for the workpiece, and the combination of the machining results as experimental machining data (step ST9).

The experimental machining condition calculation unit 21 determines whether or not to continue the experimental machining when the experimental machining data is accumulated by the actual machining result accumulation unit 24.
As a criterion for determining whether or not to continue the experimental machining, for example, the number of experiments, time, the amount of change in the machining characteristic model, the absolute value of the expected value, and the like can be considered.
If the experimental machining condition calculation unit 21 determines to continue the experimental machining (in the case of step ST11: NO), it returns to the process of step ST6 and generates the next experimental machining condition, but determines that the repetition of the experimental machining is to be stopped. Then, generation of the experimental machining conditions is stopped (in the case of YES at step ST11).

When the experimental machining condition calculation unit 21 stops generating the experimental machining conditions and the experimental machining by the processing machine 1 is completed, the machining result evaluation partial changing unit 27 stores the experimental machining data accumulated by the actual machining result accumulation unit 24. Among them, one experimental machining data is selected, and one machining quality evaluation included in the machining result in the experimental machining data is changed (step ST12).
The selection of experimental processing data may be in accordance with the order of experimental processing by the processing machine 1 or may be selected randomly.
Further, when there are N items of machining quality evaluation included in the machining results in the experimental machining data (in the example of FIG. 3, there are 4 items of machining quality evaluation), one item is selected from the N items. , Change the processing quality evaluation of the item.

Each time the machining result evaluation partial change unit 27 changes the machining quality evaluation, the machining characteristic model generation unit 28 generates a new machining characteristic model that reflects the machining quality evaluation after the change, and creates a new machining characteristic. The model is output to the machining characteristic model storage unit 29 (step ST13).
That is, the machining characteristic model generation unit 28 includes all experimental machining data including all the experimental machining data whose machining quality evaluation has been changed by the machining result evaluation partial change unit 27 (all accumulated by the actual machining result accumulation unit 24). A new machining characteristic model is generated using the experimental machining data.
The machining characteristic model generation method itself by the machining characteristic model generation unit 28 is the same as that of the machining characteristic model generation unit 25.
The machining characteristic model accumulation unit 29 accumulates a new machining characteristic model output from the machining characteristic model generation unit 28 (step ST14).
Note that the processing of steps ST12 to ST14 corresponds to trying a hypothesis that the machining quality evaluation included in the machining results in the selected experimental machining data is actually different.

Here, FIG. 7 is an explanatory diagram showing a change in the machining characteristic model accompanying a partial change in the machining quality evaluation.
In the example of FIG. 7, at the stage where the experimental machining is finished, in addition to the area surrounded by the solid line, the area surrounded by the dotted line and the solid line is also a parameter area estimated to be good machining (indicated by an arrow). The experimental machining conditions are those evaluated to be good at the stage of machining result determination).
However, if the machining quality evaluation under the experimental machining conditions indicated by the arrows is changed from good to poor and the machining characteristic model is regenerated, the condition for good machining is only the region surrounded by the solid line.

  The machining result evaluation partial changing unit 27 determines that unselected experimental machining data is left among all the experimental machining data accumulated by the actual machining result accumulation unit 24 or if unselected experimental machining data is stored. Even if it is not left, if there is a machining quality evaluation that has not been changed yet in the machining quality evaluation of the N items included in the machining result in a certain experimental machining data (step ST15: NO ), The process returns to the process of step ST12, and the process quality evaluation change process is continued.

That is, the machining result evaluation partial change unit 27 returns the previously changed machining quality evaluation to the state before the change, and the N quality machining quality evaluation of N items included in the machining result in the previously selected experimental machining data is included. Thus, the machining quality evaluation that has not been changed is changed (step ST12).
Alternatively, experimental machining data not yet selected is selected, and the machining quality evaluation included in the machining result in the experimental machining data is changed by one (step ST12).
When the machining result evaluation partial change unit 27 changes one machining quality evaluation, the machining characteristic model generation unit 28, like the above, all experimental machining data including experimental machining data in which one machining quality evaluation is changed. Is used to generate a new machining characteristic model (step ST13).
The machining characteristic model accumulation unit 29 accumulates a new machining characteristic model output from the machining characteristic model generation unit 28 (step ST14).

  The machining result evaluation partial change unit 27 does not leave unselected experimental machining data among all the experimental machining data accumulated by the actual machining result storage unit 24, and also includes all of the experimental machining data in the experimental machining data. When all the processing quality evaluations included in the processing results have been changed (step ST15: YES), the processing quality evaluation change processing is terminated.

When the processing quality evaluation changing process is completed, the machining characteristic model synthesis unit 30 synthesizes a new machining characteristic model accumulated by the machining characteristic model accumulation unit 29, and a synthesized machining characteristic model that is a synthesized machining characteristic model. Is output to the composite processing characteristic model storage unit 31 (step ST16).
Specifically, from the new machining characteristic model accumulated by the machining characteristic model accumulating unit 29, only the machining conditions estimated to obtain good machining results are extracted, and the estimated model including the machining conditions is extracted. Is a composite processing characteristic model.
Alternatively, a new machining characteristic model is synthesized by taking the intersection of the sets that provide good machining results among the new machining characteristic models accumulated by the machining characteristic model accumulation unit 29.

When the machining characteristic model synthesis unit 30 outputs the synthesized machining characteristic model to the synthesized machining characteristic model storage unit 31, the optimum machining condition calculation unit 32 calculates the optimum machining condition from the synthesized machining characteristic model and sets the optimum machining condition. The data is output to the optimum machining condition storage unit 33 (step ST17).
For example, the center of the parameter area of the combination is calculated under the processing conditions included in the composite processing characteristic model (processing conditions estimated to obtain good processing results).
As a result, it is possible to obtain machining conditions that are not easily affected by secular changes in control parameters and variations in workpiece components.

  As apparent from the above, according to the first embodiment, an experimental machining condition that is a combination of a plurality of control parameters is set using a machining characteristic model that indicates the relationship between the machining condition of a workpiece to be machined and the machining result. The experimental machining condition calculation unit 21 to be generated, and the machining results of the experimental machining are collected from the processing machine 1 that experimentally processes the workpiece according to the experimental machining conditions generated by the experimental machining condition calculation unit 21, and the machining results and the above-described experimental machining are collected. Using the machining result collection unit 12 that accumulates a set of conditions as experimental machining data and the experimental machining data accumulated by the machining result collection unit 12, a new machining characteristic model is generated, and the machining characteristic model is used as an experimental machining condition. For each experimental machining data accumulated by the machining characteristic model generation unit 25 and the machining result collection unit 12 output to the calculation unit 21, machining in the experimental machining data is performed. Each time the machining result evaluation partial change unit 27 changes the machining quality evaluation included in the results one by one, and the machining result evaluation partial change unit 27 changes the machining quality evaluation, the machining quality evaluation after the change is performed. A machining characteristic model generation unit 28 that generates a new machining characteristic model that is reflected, and a machining characteristic model synthesis unit 30 that synthesizes a new machining characteristic model generated by the machining characteristic model generation unit 28 are provided to perform optimum machining. Since the condition generation unit 14 is configured to generate an optimum machining condition from the machining characteristic model synthesized by the machining characteristic model synthesis unit 30, it is possible to search for the best machining condition with a small number of experiments and to perform machining. Even if there is an error in the quality evaluation of the result, there is an effect that the influence of the error can be eliminated and an appropriate machining condition can be generated.

The effects obtained in the first embodiment are specifically as follows.
In the first embodiment, every time experimental machining is performed, machining results are additionally accumulated as data in the actual machining result accumulation unit 24, and a new machining characteristic model is generated based on the data. Then, the experimental machining condition calculation unit 21 uses the new machining characteristic model to generate a machining condition that is expected to be the most effective in knowing the machining characteristics as the next experimental machining condition. Therefore, when determining the machining conditions corresponding to the required specifications, the desired machining conditions can be easily searched with a small number of experiments.

  In the first embodiment, the machining result evaluation partial changing unit 27 changes the machining quality evaluation included in the machining results in the experimental machining data accumulated by the actual machining result accumulation unit 24 one by one. The machining characteristic model generation unit 28 generates a new machining characteristic model (machining characteristic model when the machining result evaluation is incorrect) reflecting the machining quality evaluation after the change, and the machining characteristic model synthesis unit 30. However, the new machining characteristic model generated by the machining characteristic model generation unit 28 is synthesized, and the optimum machining condition generation unit 14 generates an optimum machining condition from the machining characteristic model synthesized by the machining characteristic model synthesis unit 30. ing. Therefore, the optimum condition can be selected from among the machining conditions that are not easily affected by the judgment error of the machining result, and the frequency at which an undesirable machining condition is output for the machining that may cause the judgment of the machining result to be erroneous Decrease.

By the way, in the first embodiment, generation of a machining characteristic model (step ST6), generation of experimental machining conditions (step ST7), execution of experimental machining (step ST8), judgment of machining results (step ST9), and experimental machining data (Step ST10) are repeatedly executed in order, and the experimental machining condition calculation unit 21 determines whether to continue or stop the generation of the experimental machining conditions according to the number of experimental machinings and the amount of change in the machining characteristic model. .
Further, in the first embodiment, partial modification of the machining result evaluation (step ST12), generation of the machining characteristic model (step ST13), and accumulation of the machining characteristic model (step ST14) are repeatedly performed in order. The result evaluation partial change unit 27 determines whether to continue or stop the partial change of the machining result evaluation.
Therefore, it is possible to automatically repeat the experimental machining. At this time, by performing the experimental machining for accelerating the optimization of the machining characteristic model, the time until obtaining a better machining condition as a whole is shortened.

In the first embodiment, the experimental machining condition calculation unit 21 generates one experimental machining condition at step ST7 for each period of step ST7. However, a plurality of experimental machining conditions are generated each time, The processing machine 1 may perform a plurality of experimental processings each time according to a plurality of experimental processing conditions.
In experimental machining that requires setup work by an operator, it may be desired to perform a plurality of experimental machinings in a single stage. In such a case, by configuring the experimental machining condition calculation unit 21 to generate a plurality of experimental machining conditions for each period of step ST7, the number of such setup changes can be reduced.

In the first embodiment, the case where the processing condition parameters are two parameters a and b is shown, but the number of processing condition parameters may be two or more (for example, five or seven). The effect of can be obtained.
As the number of parameters increases, the number of combinations becomes enormous. Therefore, when the number of parameters increases, the present invention that can reduce the number of experiments is more advantageous.
When there are a plurality of processing condition parameters, the above-described area calculation part may be replaced with an amount calculation corresponding to the volume of the manifold in the multidimensional space.
Alternatively, the difference between the current estimated model and the expected estimated model may be calculated using the calculation formula of the Cullback library divergence, and the difference may be used.

Embodiment 2. FIG.
In the first embodiment, the processing result determination unit 23 uses the automatic measuring device to quantify the processing result output from the processing machine 1 and evaluates the quality of the processing, but the processing result determination unit 23 may be provided with a man-machine interface that displays the processing result output from the processing machine 1 and receives the setting of the quality evaluation of the processing.
For example, when it is very expensive to automate the measurement, or when there is no measurement value suitable for determining the machining result, manual measurement or judgment by appearance is required.
Therefore, in the second embodiment, the processing result determination unit 23 is configured to include a man-machine interface.

The processing result determination unit 23 is provided with an operator interface screen capable of displaying experimental processing conditions and measured numerical data and inputting a determination result by appearance. That is, the processing result determination unit 23 is set as one screen of an NC (numerical control unit) device of the processing machine 1.
Here, the measured numerical data is numerical data obtained by manually measuring a workpiece obtained by experimental machining.
Unlike the above, a configuration in which a similar operator interface screen is provided on a separate computer connected to the processing machine 1 via a network may be adopted.

For example, a mode is conceivable in which the user manually measures the machining result shape of the workpiece and inputs the measurement result as numerical data on the operator interface screen.
Here, FIG. 8 is an explanatory diagram showing an example of the operator interface screen.
Hereinafter, the operation of the machining condition search apparatus according to the second embodiment will be described with reference to FIG.

For example, if the experimental machining condition calculation unit 21 generates four initial experimental machining conditions, four initial experimental machining conditions are displayed on the operator interface screen as shown in FIG.
In this case, the processing machine 1 performs the experimental processing four times according to the four initial experimental processing conditions.
When the four experimental machinings are completed, the operator manually measures or determines items such as the surface roughness of the workpiece, the presence / absence of dross, and the possibility of cutting, and inputs the results.
It should be noted that the experimental machining based on the initial experimental machining conditions, the measurement / judgment of the machining results by manual work, and the input process of the measurement / judgment results may be sequentially repeated for each initial experimental machining condition.
The measurement data is numerical data, and the determination result is binary numerical data such as “1” or “−1”, or a designated character string or symbol. When inputting a character string or a symbol, the actual machining result storage unit 24 converts the character string or symbol into binary numerical data such as “1” or “−1”.

Next, the machining characteristic model generation unit 25 generates a machining characteristic model using the initial experimental machining conditions and the input measurement values.
Thereafter, the experimental machining condition calculation unit 21 newly generates, for example, three experimental machining conditions using the machining characteristic model.
Thereafter, the processing machine 1 performs experimental processing according to the newly generated experimental processing conditions.
Next, the operator measures and determines the processing result generated by the processing machine 1, and inputs the measurement and determination result on the operator interface screen shown in FIG.

  In addition, after all of the experimental machining based on the three experimental machining conditions is completed, measurement / judgment and input processing of a workpiece manufactured by the experimental machining may be performed. In addition, after the experimental machining based on one or two experimental machining conditions is completed, the experimental machining is temporarily stopped, and measurement / judgment of the workpiece produced by the experimental machining is performed, and then input processing is performed. The experimental machining based on the remaining experimental machining conditions may be resumed, and measurement / judgment and input processing may be performed on the workpiece produced by the experimental machining.

When the dimension measurement result is manually input on the operator interface screen, the input numerical data is immediately stored in the actual machining result storage unit 24.
Then, the machining characteristic model generation unit 25 newly generates a machining characteristic model by using the numerical data accumulated in the actual machining result accumulation unit 24 and each data already stored.
Next, as described above, when the machining result measurement result and the determination result are manually input to the operator interface screen, the experimental machining condition calculation unit 21 generates the machining generated by the machining characteristic model generation unit 25. For example, three experimental processing conditions are newly generated using the characteristic model.

In any stage, it is assumed that the operator inputs “experimental processing conditions” displayed on the operator interface screen. In this case, among the experimental processing conditions that have already occurred, the remaining experimental processing conditions that have not yet been experimentally processed are deleted.
The experimental machining condition calculation unit 21 newly generates, for example, three experimental machining conditions using the machining characteristic model at the time. Then, the processing machine 1 automatically performs experimental processing based on the newly generated experimental processing conditions.
In addition, when the experimental machining condition calculation unit 21 determines the end of the experimental machining or when “experimental machining end” displayed on the operator interface screen is input, the execution of the new experimental machining is stopped. Then, the machining result evaluation partial changing unit 27 changes the actual machining result evaluation. Then, the optimum machining condition calculation unit 32 generates optimum machining conditions.

  As described above, according to the second embodiment, since the operator interface screen for inputting the measurement result and the determination result for the workpiece from the outside is provided, manual processing for the workpiece manufactured by the processing machine 1 is provided. It is possible to provide a machining condition searching apparatus that can cope with measurement and when determination by a person is necessary.

Embodiment 3 FIG.
In the first and second embodiments, the machining result evaluation partial changing unit 27 evaluates the machining quality evaluation included in the machining results in all the experimental machining data accumulated by the actual machining result accumulation unit 24 one by one. Although what is to be changed is shown, a part of the processing results whose processing quality evaluation is changed are selected from the processing results in the experimental processing data accumulated by the actual processing result storage unit 24 and included in the processing results. You may make it change the process quality evaluation currently performed one by one.

FIG. 9 is an explanatory diagram showing a method of selecting a processing result for changing the processing quality evaluation.
The machining result evaluation partial change unit 27 changes the machining quality evaluation included in some machining results among the machining results in the experimental machining data accumulated by the actual machining result accumulation unit 24. For example, among the processing results determined to be good, the processing results of the adjacent experimental processing that are determined to be bad are set as processing result candidates for changing the processing quality evaluation.
The machining result evaluation partial change unit 27 selects one machining result from the machining result candidates whose machining quality evaluation is to be changed, and changes the machining quality evaluation included in the machining result.

When the machining result evaluation partial change unit 27 changes the machining quality evaluation, the machining characteristic model generation unit 28, like the first embodiment, all the experimental machining data including the experimental machining data in which the machining quality evaluation is changed. Is used to generate a new machining characteristic model, and the machining characteristic model is stored in the machining characteristic model storage unit 26.
The machining result evaluation partial changing unit 27 returns the changed machining quality evaluation to the state before the change, and repeatedly performs the above-described processing until the change of the machining quality evaluation is completed for all candidates.

Regarding the reliability of the determination of the processing result, not all the determinations are low in reliability, but there are a determination result with high reliability and a determination result with low reliability.
In the third embodiment, evaluations of two adjacent machining results are compared, and when two evaluations are different, slight conditions (aging of workpiece components and control parameters, humidity, temperature) We focus on the fact that the evaluation of the machining result may have changed due to the change, and it may be difficult to make a judgment by appearance, and the reliability of the judgment of the machining result is low.
In addition, as a selection criterion for the processing result candidate whose evaluation is to be changed, for example, the processing result candidate may be within a certain distance from the boundary drawn by a solid line, or Mth (for example, 10th in order of increasing distance from the boundary). )).
According to the third embodiment, since the number of machining results whose evaluation is changed can be reduced, there is an effect that the time required for searching for the optimum machining conditions can be reduced.

  In the present invention, within the scope of the invention, any combination of the embodiments, or any modification of any component in each embodiment, or omission of any component in each embodiment is possible. .

  DESCRIPTION OF SYMBOLS 1 Processing machine, 11 Experimental machining condition production | generation part, 12 Machining result collection part (machining result collection means), 13 Machining characteristic model change part, 14 Optimal machining condition production | generation part (optimum machining condition production | generation means), 21 Experimental machining condition calculation part (Experimental machining condition generation unit), 22 Experimental machining condition storage unit, 23 Machining result determination unit, 24 Actual machining result storage unit, 25 Machining characteristic model generation unit (first machining characteristic model generation unit), 26 Machining characteristic model storage 27, Machining result evaluation partial change unit (second machining characteristic model generation means), 28 Machining characteristic model generation unit (second machining characteristic model generation means), 29 Machining characteristic model storage unit, 30 Machining characteristic model synthesis Unit (machining characteristic model synthesis means), 31 synthetic machining characteristic model storage unit, 32 optimum machining condition calculation unit, 33 optimum machining condition storage unit.

Claims (4)

An experimental machining condition generating means for generating an experimental machining condition that is a combination of a plurality of control parameters using a machining characteristic model indicating a relationship between a machining condition of a workpiece to be machined and a machining result, and the experimental machining condition generating means Machining result collection means for collecting the machining results of the experimental machining from a processing machine for experimentally machining the workpiece according to the generated experimental machining conditions, and storing a set of the machining results and the experimental machining conditions as experimental machining data; and First machining characteristic model generation means for generating a new machining characteristic model using the experimental machining data accumulated by the machining result collecting means, and outputting the machining characteristic model to the experimental machining condition generation means, and the machining For each experimental machining data accumulated by the result collection means, one machining quality evaluation included in the machining results in the experimental machining data From good to bad, or while changing from bad to good, and the second processing characteristics model generation means for generating a new processing properties model processing quality evaluation after the change is reflected, the second processing characteristics model Machining characteristic model synthesis means for synthesizing a new machining characteristic model generated by the generation means, and optimum machining condition generation means for generating an optimum machining condition from the machining characteristic model synthesized by the machining characteristic model synthesis means. Machining condition search device. The processing result collecting means, by digitizing the processed result output from the processing machine, provided with a processing result judging unit that evaluates the quality of machining, the machining quality evaluation is the determination result of the processing result determining section comprising processing results and processing conditions search device according to claim 1, wherein a set of the above experiments processing conditions, characterized in that accumulate as the experimental processing data. The processing result collecting unit displays a processing result output from the processing machine, equipped with a man-machine interface for receiving settings of quality evaluation processing, the processing quality evaluation set by the man-machine interface has been accepted processing results and processing conditions searching apparatus according to claim 1 or claim 2, wherein a set of the above experiments processing conditions, characterized in that accumulate as the experimental machining data including. It said second processing characteristics model generation unit from among the processing results of accumulated within experimental machining data by the processing result acquisition means, selecting a part of the processing result of changing the machining quality evaluation, the processing result The processing quality evaluation included in the process is changed from good to defective one by one or from defective to good, and a new processing characteristic model reflecting the changed processing quality evaluation is generated. The processing condition search device according to any one of claims 1 to 3.

Images