JPH04350702A - Method for optimizing set condition of machine tool - Google Patents

Abstract

PURPOSE:To accurately and easily reduce a defect of working of a machine tool and improve the quality of working by using fuzzy multiple regression to calculate an optimum condition. CONSTITUTION:A calculation condition and a set condition are inputted from a keyboard 5 and are sent to a fuzzy multiple regression and optimization calculating device 3 through a controller 2. A machine tool 1 is driven in accordance with the set condition to perform working. The quality of a worked article in the set condition is visually evaluated by a person; and if the number of data does not satisfy a certain condition, the set condition is inputted again from the keyboard 5 to perform visual evaluation. When the number of data satisfies the condition, fuzzy multiple regression analysis is performed by the fuzzy multiple regression and optimization calculating device 3 to display the optimum condition on a CRT 4 through the controller 2.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for optimizing processing machine settings, which is applied to processing machines such as plastic molding machines.

0002

[Prior Art] Generally, in processing machines such as plastic molding machines, there is a close relationship between machine setting conditions and processing quality, and machine setting conditions have a large influence on obtaining good processing quality. It is known. As a method for optimizing processing quality, the present applicant previously proposed a patent application filed in 1983.
No. 0-90424 (Japanese Unexamined Patent Publication No. 61-248723) entitled ``Method for optimizing setting conditions of processing machines.'' The processing machine setting condition optimization method in this application evaluates the processing quality as shown in the flowchart of FIG. The optimal conditions are calculated from the regression equation. This method is based on the premise that machining quality is actually measured numerically due to multiple regression. However, most machining quality is difficult to measure and is often ranked visually. The rank value of this visual evaluation is not absolute, and when re-evaluated, it often deviates up or down from the original rank, and has ambiguity. Therefore, if the rank value of the visual evaluation is used as a normal numerical value and multiple regression is performed, the ambiguity will be ignored, and the result will be different from the actual situation.

0003

[Problems to be Solved by the Invention] As mentioned above, in the conventional technology, when processing quality is visually evaluated by ranking, multiple regression is performed in order to ignore the ambiguity of the ranking evaluation. Even if an attempt was made to utilize this to optimize the setting conditions for processing machine operation, there was a problem in that the results would differ from the actual situation.

The present invention solves the above-mentioned problems, and even when machining quality is visually ranked and evaluated, the relationship between machining quality and setting conditions can be accurately grasped, and the setting conditions can be optimized to suit the actual situation. The purpose is to provide a method for

[0005]

[Means for Solving the Problems] The present invention treats the machining quality rank value of visual evaluation as a fuzzy value, performs fuzzy multiple regression with the setting condition factors of processing machine operation, and solves the MAX problem of fuzzy multiple regression. This is a system that calculates and sets setting conditions that optimize machining quality using a multiple regression equation that shows the average of the upper and lower limit solutions of , and uses this as a means of problem solving.

[0006]

[Operation] When machining quality is visually evaluated by ranking, multiple regression is performed by treating the machining quality rank value as a fuzzy value, so multiple regression is performed taking into account the ambiguity of the ranking evaluation. In addition, in the MAX problem of fuzzy multiple regression, two solutions, an upper limit solution and a lower limit solution, are obtained, so the relationship with the setting condition factors cannot be clarified as is. By using a multiple regression equation that shows, the relationship between machining quality and setting condition factors is uniquely determined, so the setting conditions that optimize the machining quality can be determined from the multiple regression equation. Therefore, it is possible to optimize the setting conditions of the processing machine to suit the actual situation, taking into account the ambiguity of visual ranking evaluation.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. Embodiments of the present invention are shown in FIGS. 1 to 3.

FIG. 1 shows the system configuration of this embodiment, where 1 is a processing machine such as a plastic molding machine, 2 is a control device, 3 is a fuzzy multiple regression/optimization calculation device, and 4 is a CRT.
(Cathode Ray Tube is a so-called cathode ray tube), and 5 is a keyboard. Further, 6 is a user interface, which connects the control device 2, CRT 4, and keyboard 5. Note that numerals 11 to 15 are signal lines connecting the respective elements.

The control device 2 sends operating setting conditions to the processing machine 1 and receives information on the actual operating state. The control device 2 also has a user interface 6.
Display information is sent to the CRT 4 via the CRT 4, and input information from the keyboard 5 is received. Furthermore, the control device 2 sends information on the setting conditions and operating conditions of the processing machine 1 and input information from the keyboard 5 to the fuzzy multiple regression/optimization calculation device 3, and conversely sends the calculation results of the calculation device 3. receive.

FIG. 2 shows a flowchart of optimization of processing machine setting conditions by this system. In the same figure, in step 26, the number n of setting condition factors to be optimized, the name of each of the n setting condition factors, and the minimum and maximum values of their change levels are manually input from the keyboard 5 of FIG. It is sent via the control device 2 to the fuzzy multiple regression/optimization calculation device 3. Next, in step 27, the level of the setting conditions during machining is transmitted to the calculation device 3 from the keyboard 5 or the control device 2 where the conditions have been set. In this case, it is also possible to send the setting conditions input from the keyboard 5 to the control device 2 to set the conditions for the processing machine 1.

Further, in step 28, processing is performed by driving the processing machine 1 via the control device 2 in accordance with the above set conditions. Then, in step 29, a human visually evaluates the quality of the processed product under the above-mentioned setting conditions and performs an evaluation such as rank 3 on a five-level evaluation.

[0012] Next, in step 30, it is determined whether or not the number of data of the visual evaluation value of the processing quality has been collected enough to perform the fuzzy multiple regression analysis described later. A data shortage signal is output, and steps 27 to 30 are repeated to collect the data for the shortage. In this case, the above-mentioned number of data refers to the number of sets where one set (n pieces) of setting conditions and the machining quality visual evaluation value rank under that setting condition are set as one set,
To perform fuzzy multiple regression, which will be described later, the number of data sets (n+1) or more is required.

In the next step 31, the machining quality visual evaluation value rank Y is converted into a characteristic value and each setting condition factor x1, x2
,...,xn as explanatory variables, fuzzy multiple regression analysis is performed. Here, the machining quality visual evaluation value rank Y is treated as a fuzzy value. For example, rank 3 is displayed as (center 3, width 1), and can take rank values from 2 to 4. Each setting condition factor that is an explanatory variable is a crisp value (normal value that is not a fuzzy value).

In the fuzzy multiple regression analysis in step 31, fuzzy multiple regression is first performed to find a solution to the MAX problem, which will be described later, and then the average solution to the MAX problem is calculated. Then, the average solution of this MAX problem is solved, and in the next step 32, optimal conditions are calculated and displayed. Furthermore, in the next step 33, it is determined whether machining can be continued based on this optimal condition, and if it is possible, the machining is repeated from step 27 described above.

There are two types of fuzzy multiple regression problems: MIN problem and MAX problem. MIN problem is to find the minimum width that includes all the input data, and MAX problem is to find the minimum width that includes all the input data. It is the widest one that passes through. These are formulated as follows. [Model] Y(X)=A0 +A1 x1 +
A2 x2 +…+An xn However, Y: Visual evaluation rank
fuzzy value
Ai: Coefficient Ai = (center aci, width awi) Fuzzy value
X: Forming factor X=(x1,
x2 ,...,xn) fuzzy value
xi: each factor
Crisp value (ordinary numeric variable) Therefore, Y
(X) = (center yc (X), width yw (X))
yc (X)=ac0+a
c1x1 +ac2x2 +...+acnxn
yw (X)=aw0+aw
1x1 +aw2x2 +...+awnxn [Input data] Xj = (x1j, x2j,..., xnj),
Yj = (center ycj, width yw
j) However, j = 1, 2, ..., m (number of sets) [MIN problem] Constraint condition: yc (Xj ) - yw
(Xj)≦ycj−ywj
yc (Xj)+yw
(Xj)≧ycj+ywj

j=1, 2,...,m Objective function: yw (X1)+yw (X2)+...+y
w (Xm) Minimum [MAX problem] Constraint condition: yc (
Xj )-yw (Xj)≧ycj-ywj
yc (X
j )+yw (Xj)≦ycj+ywj

j=1, 2,..., m
Objective function: yw (X1) + yw (
X2 ) +...+yw (Xm) Maximum [Solution method] Solve the MIN problem and MAX problem using linear programming.

[0016] Processing quality visual evaluation value rank Y is determined by 2 factors x 1
, x2 is shown in FIG. 3. However, of the two factors, x2 = C (constant). In the same figure, the upper and lower bound solutions of the MIN problem are (
1) and (4), and similarly, the upper and lower limit solutions of the MAX problem are (2) and (3). As can be seen from the figure, MI
Solutions (1) and (4) for the N problem have little utility value because the range between them is too wide. On the other hand, solution (2) of the MAX problem, (
In 3), the width between them is a reasonable width when viewed from the original fuzzy value data Y, and can be put to practical use. Note that the normal multiple regression B is considered inappropriate because it is a single line and the width of the original fuzzy value data Y is not considered at all. As mentioned above, solutions (2) and (3) for the MAX problem are considered appropriate, but in this example, (2) is a straight line that slopes upward to the right, and (3)
Since is a straight line that slopes downward to the right, it cannot be determined whether factor x1 should be increased or decreased in order to improve the processing quality visual evaluation value rank Y. Therefore, both (2) and (3)
We created a new average (5) of , and devised a way to find the optimal conditions based on the tendency of (5). In this figure, since (5) slopes slightly to the right, the smaller the value of factor x1, the larger Y becomes. Therefore, if the maximum and minimum values of factor x1 are given, Y will be maximized if x1 is set to the minimum value. Similarly, by examining increases and decreases in the processing quality visual evaluation value rank Y, it is possible to determine the values of each factor that maximize Y, that is, the optimal conditions. Although the case where the fuzzy multiple regression equation is a linear equation has been described above, it may be a quadratic equation as follows.

[0017]

[Math 1]

Alternatively, xi xj , xi /xj
A model including multiple factors such as the following may also be used. 1 of these
In models other than the following formula, the average formula for the MAX problem of fuzzy multiple regression is not a linear formula, so the maximum or minimum value of each explanatory variable xi is not necessarily the optimal condition.
Optimal conditions can be calculated from the above average formula using general mathematical techniques.

[0019]

[Effects of the Invention] As described in detail above, according to the present invention, when visually evaluating machining quality by ranking, the fuzzy multiple regression in which the rank value is a fuzzy value, and the upper and lower limit solutions of the MAX problem are used. By calculating the optimal conditions using the average formula, it is possible to optimize the processing machine settings to suit the actual situation, taking into account the ambiguity of visual ranking evaluation. There has been no technology in the past to optimize the setting conditions of processing machines by treating such visual evaluation rank values as fuzzy values, so by putting into practice the setting condition optimization of the present invention, it is possible to reduce processing defects in processing machines. Processing quality can be improved more accurately and easily than before.

[Brief explanation of drawings]

FIG. 1 is a system configuration diagram showing an embodiment of the present invention.

FIG. 2 is a flowchart of a method for optimizing setting conditions of a processing machine using the same system.

FIG. 3 is a diagram showing the results of fuzzy multiple regression.

FIG. 4 is a flowchart showing a conventional method for optimizing setting conditions of a processing machine.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1...Processing machine, 2...Control device, 3...Fuzzy multiple regression/optimization calculation device, 4...CRT, 5...Keyboard, 6...User interface, 1-15...Signal line, 26-33...
Each process of the system.

Claims (1)

[Claims] 1. A first step of setting a change level of a setting condition factor of a processing machine; a second step of setting a condition of the processing machine according to a change level of the setting condition factor;
a third step in which the processing machine is driven to perform processing according to set conditions; a fourth step in which the quality of the processed product processed in this third step is visually ranked and evaluated; In the third step, it is determined whether the number of data of the ranked machining quality evaluation values is sufficient to perform the fuzzy multiple regression analysis, and if the number is not, a data number shortage signal is output to the second step. A fifth step of collecting a predetermined amount of data, and in a state where this fifth step is possible, the fuzzy multiple regression analysis is performed using the machining quality evaluation value rank as a characteristic value and the setting condition factor as an explanatory variable. The sixth step is to find the upper and lower limit solutions of the MAX problem and calculate the average solution.
process and the fuzzy multiple regression M obtained in this sixth process.
A seventh step of calculating and displaying the setting conditions that optimize the forming quality based on the tendency of the average solution of the upper limit solution and the lower limit solution of the AX problem, and determining whether or not to continue processing based on the optimum value of the setting condition factor. Setting conditions for a processing machine characterized by comprising: an eighth step of outputting a machining continuation signal to the second step in a possible state to cause the machining to continue under the set conditions, and stopping the machining in a negative state. Optimization method.