JPH06198544A - Feedback type machining condition correction device - Google Patents

Abstract

PURPOSE:To improve a correction precision of a machining condition by correcting the machining condition according to a regulation which changes accompanying an oscillatory level, which is an oscillatory fluctuating level of a measurement value by a measurement device. CONSTITUTION:In a feedback type fixed size position correction device, which enumerates an interim correction value U for an interim value of a correction value U on the basis of a measurement value X according to fuzzy inference and decides the final correction value U* for the final value of the correction value U in this time on the basis of the interim correction value U, the interim correction value U is enumerated using a negative (dull) fuzzy rule (S304) when an oscillatory level of the measurement value X is high and an abnormal flag is ON (S301, 302). On the other hand, a positive (sharp) fuzzy rule is used to enumerate the interim correction value U (S303) when an osciilatory level is low and the abnormal flag is off (S301, 302).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention is designed to feed back information regarding the dimensional error of a machined workpiece,
The present invention relates to a feedback type machining condition correction device for correcting the machining conditions of a work to be machined next.

[0002]

2. Description of the Related Art The present applicant is studying the following feedback type machining condition correcting device. that is,
(a) A processing machine that sequentially processes each of a plurality of workpieces, and (b)
Processing conditions of the processing machine are determined based on a correction value supplied from the outside, and processing machine control means for controlling the processing machine according to the determined processing conditions, and (c) a plurality of processing machines processed by the processing machine. A feedback type machining condition correcting device to be used by being connected to the machining machine control means and the measuring machine of the machining system including a measuring machine that measures each dimension of the work piece in order, and (d) the measuring machine Correction value determining means for determining the correction value of the processing condition of the work to be processed next by the processing machine based on the measured value according to, and (e) the correction value for supplying the determined correction value to the processing machine control means. And a supply means.

The applicant of the present invention implemented this feedback type machining condition correction device in the following manner. That is, the correction value is determined according to the only rule based on the measured value.

[0004]

Normally, the measured value increases the number of works measured by the measuring machine (hereinafter referred to as "measured work number") due to wear of the working tool of the working machine. Changes gradually with. In such a state where the measured value is stable, it is desirable to determine the correction value so as to respond sensitively to the error value which is the difference between the measured value and the target value of the machining dimension of the workpiece. The only rule above is predetermined to satisfy

However, the measured value may periodically fluctuate as the number of workpieces to be measured increases due to vibration of the processing machine or the like. In this way, in the state where the measured value oscillates, that is, in the unstable state, unlike the above-mentioned stable state, when the correction value is determined so as to respond sensitively to the error value, rather than the measured value becoming stable, On the contrary, the vibration level of the measured value may increase.

In summary, when the correction value determination rule is fixed, it is difficult to secure the machining accuracy of the workpiece regardless of whether the measured value is in a stable state or an unstable state. There is.

Against the background of such circumstances, the invention of claim 1 has been made to solve the above problems by making the correction value determination rule variable.

The invention of claim 2 is an aspect of use of the invention of claim 1, in which the decision rule can be changed by selecting an appropriate one from a plurality of decision rules prepared in advance. It is an object of the invention to provide an aspect to be realized.

Further, the invention of claim 3 is another utilization mode of the invention of claim 1 in which the decision rule is changed by appropriately changing the only decision rule prepared in advance. It was made as an issue to provide.

[0010]

In order to solve each of the problems, the invention of claim 1 is, as shown in FIG. 1, a processing machine equipped with the processing machine 1, processing machine control means 2 and measuring machine 3. A feedback type machining condition correction device which is to be used by being connected to the processing machine control means 2 and the measuring machine 3 of the system, and which includes the correction value determination means 4 and the correction value supply means 5 It is characterized in that the deciding means 4 decides the correction value based on the value measured by the measuring machine 3 according to a rule that changes with the vibration level which is the vibrational fluctuation level of the measured value.

The "correction value determining means 4" here
Is a mode in which the correction value is determined only based on an error value that is the difference between the measured value and the target value of the machining dimension of the work, or not only the error value but also its change tendency (the error value is
The correction value is also determined based on the tendency of change as the number of measured works increases, or the correction value is determined not only based on those error values and their change tendency, but also on the change tendency of the change tendency. It can be an aspect. Further, the true value of the current measured value is estimated based on the current measured value and the past measured value, and the estimated true value is regarded as the actual measured value to determine the correction value. You can also

According to a second aspect of the present invention, the correction value determining means 4 according to the first aspect of the invention measures the vibration level of the measured value and matches the measured vibration level from a plurality of preset rules. It is characterized in that the selected value is selected and the correction value is determined according to the selected rule.

According to a third aspect of the present invention, the correction value determining means 4 according to the first aspect of the present invention measures the vibration level of the measured value and corrects a preset rule based on the measured vibration level, It is characterized in that the correction value is determined according to the corrected rule.

[0014]

In the feedback type machining condition correcting apparatus according to each of the first to third aspects of the invention, the correction value determining means 4 corrects based on the value measured by the measuring machine 3 according to a rule that changes with the vibration level of the measured value. The value is determined. The correction value is determined according to a rule suitable for the vibration level of the measured value. For example, when the vibration level is high, the rule is that the correction value is determined so that it responds to the dimensional error of the work insensitively, while when the vibration level is low, the dimensional error of the work is reduced. The rule is such that the correction value is determined so as to respond sensitively.

In particular, in the feedback type machining condition correcting apparatus according to the second aspect of the present invention, the vibration level of the measured value is measured by the correction value determining means 4, and the vibration level is measured from a plurality of preset rules. The one that matches the vibration level is selected, and the correction value is determined according to the selected rule.

Further, in particular, in the feedback type machining condition correcting apparatus according to the invention of claim 3, the correction value determining means 4 measures the vibration level of the measured value, and the vibration level of the measured value is set in advance based on the measured vibration level. The corrected rule is corrected, and the correction value is determined according to the corrected rule.

When carrying out the invention of claim 1,
The invention of claim 2 and the invention of claim 3 can be embodied at the same time. That is, for example, from a plurality of preset rules, select one that matches the measured vibration level, based on the measured vibration level,
It can be implemented in a manner to correct the selected rule.

[0018]

As is clear from the above description, according to the inventions of claims 1 to 3, the correction value is determined according to an appropriate rule in relation to the vibration level of the measured value.
It is possible to obtain the effect that the dimensional accuracy of the work is ensured regardless of the vibration level of the measured value.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A feedback type fixed point correction device, which is an embodiment common to the inventions of claims 1 to 3, will be described in detail below with reference to the drawings.

This sizing point correction device is used in conjunction with a machining system in which a crankshaft of an automobile engine is used as a workpiece to be machined, and a plurality of journal surfaces formed in advance are cylindrically ground as machining sites. As shown in FIG. 2, the crankshaft is a work having journal surfaces that are seven outer peripheral cylindrical surfaces (hereinafter, simply referred to as “cylindrical surfaces”) arranged coaxially with each other.

Specifically, as shown in FIG. 3, the processing system includes a processing line, a processing machine 10, two in-process measuring machines 12 (shown as one in the figure), a sizing device 14,
It is composed of a motor controller 15, a total number measuring machine 16, a work number counter 18, a control device 20, an auxiliary storage device 22 and the like, and these elements will be individually described below.

The processing line is represented by a thick solid line with an arrow in the figure, and a plurality of works are conveyed in a line from the upstream side to the downstream side (from the left side to the right side in the figure). Is.

The processing machine 10 performs cylindrical grinding on each of the seven journal surfaces of the crankshaft with a circular grindstone as a processing tool. Specifically, as shown in FIG. 4, a grindstone group 3 in which a plurality of grindstones are coaxially arranged
By rotating 0 and the crankshaft in contact,
It is a multi-grinding machine that simultaneously performs cylindrical grinding on all seven journal surfaces. The configuration will be briefly described below.

The processing machine 10 has a work table 32 for a work. The work table 32 is attached to a main frame (not shown) of the processing machine 10.
The work table 32 is provided with a holding device (not shown) that holds the work rotatably around its axis and a work motor 34 that rotates the held work.

The processing machine 10 further includes a forward / backward table 36 for the grindstone group 30 and a swing table 38. The forward / backward table 36 is attached to the main frame in a state capable of reciprocating motion in a direction perpendicular to the work held by the work table 32. On the other hand, the swing table 38 has a swing axis line (a straight line extending in a direction perpendicular to the paper surface in the figure) set on the forward / backward table 36 in a state orthogonal to the axis of the grindstone (indicated by a dashed line in the figure).
Swing centering around is possible (right rotation or left rotation is possible)
It is installed in a normal state. Forward / backward table 36
The forward / backward movement of the motor is fixed to the main frame.
0, the swing of the swing table 38 is realized by the swing motor 42 fixed to the forward / backward table 36.

That is, in the processing machine 10, the angle formed by the grindstone axis and the rotation axis of the workpiece (hereinafter referred to as "cutting angle") can be adjusted by the swing motor 42.

The two in-process measuring machines 12 are attached to the processing machine 10. As shown in FIG. 2, each of the in-process measuring machines 12 has a pair of measuring elements sandwiching one cylindrical surface from both sides of the outer circumference, and measures the diameter of the cylindrical surface by an electric micrometer method. . The in-process measuring machines 12 are not individually prepared for seven journal surfaces, but as shown in the figure, the journal surfaces at both ends, that is, the first journal surface and the seventh journal surface (hereinafter, (Also referred to as “two end cylindrical surfaces”).

The sizing device 14 is, as shown in FIG.
Each of them is connected to the in-process measuring machine 12. The sizing device 14 is mainly composed of a computer including a CPU, a ROM, a RAM, and a bus, and during grinding by the processing machine 10, the diameters of the two end cylindrical surfaces are passed through the respective in-process measuring machines 12. When the remaining depth of cut on each end cylindrical surface (the amount that must be cut before reaching the final dimension) reaches each set value (existing for each end cylindrical surface), a signal to that effect. (Hereinafter, "set amount arrival signal"), when reaching each final dimension, that is, each sizing point (existing for each end cylindrical surface), a signal to that effect (hereinafter, "sizing point arrival signal") ) Is output to the motor controller 15 in association with each end cylindrical surface.

The sizing device 14 is also designed so that each sizing point can be corrected. Specifically, when each correction value U (existing for each end cylindrical surface) is supplied from the control device 20, each correction value U is added to each current sizing point to obtain each current value. It is designed to change the sizing points and keep each current sizing point if it is not supplied. That is, in the sizing device 14, the sizing point is automatically corrected by the control device 20.

As shown in FIG. 3, a keyboard 50 is connected to the sizing device 14, and when the keyboard 50 is operated by an operator, the sizing device 14 operates a manual correction value according to the operation. It is also designed to make manual corrections that only change the current sizing point. The sizing device 14 also
The latest manual correction value and the sizing point are stored in their own RAMs and voluntarily transmitted to the control device 20.
However, since the control device 20 does not always receive the latest manual correction value and the current sizing point from the sizing device 14, the sizing device 14 transmits only when the control device 20 permits reception. Will be done.

The motor controller 15 is connected to the sizing device 14, the forward / backward motor 40, etc., as shown in FIG. The motor controller 15 controls the forward / backward motor 40 and the like based on a command from an operator, a signal from the sizing device 14, and the like.

By the way, the processing machine 10 completes one round of cylindrical grinding through several steps such as rough grinding, fine grinding, spark out, etc. in order. The rough polishing is performed until the remaining depth of cut reaches the set amount, and the fine polishing is performed until the diameter reaches the sizing point. It is usual that the two set amount arrival signals to be supplied from the sizing device 14 for each end cylindrical surface do not coincide in the supply timing, and the motor controller 1
In the rough grinding stage, 5 controls the forward / backward motor 40 and the swing motor 42 in accordance with the amount of discrepancy in the signal supply timing, thereby appropriately controlling the cutting angle. Further, in the fine grinding, since the cutting angle should be appropriate in the rough grinding preceding that, the motor controller 15 drives only the forward / backward motor 40 to cut the grindstone group 30 into the work. Continue to 2
If a sizing point arrival signal is supplied to any one of the end cylindrical surfaces, the forward / backward motor 40 is stopped, spark-out is performed, and then the forward / backward motor 40 is rotated in the reverse direction to grind stone groups. 30 is retracted from the work. It is also possible to control the cutting angle even at the stage of fine polishing.

As shown in FIG. 3, the total number measuring machine 16 is arranged downstream of the processing machine 10 on the processing line. The total number measuring machine 16 has the same number of post-process measuring machines 44 as the number of cylindrical surfaces in one work, and by the same method as the in-process measuring machine 12, all the work carried out from the processing machine 10 are sequentially processed. The diameter is measured individually on all cylindrical surfaces. The total number measuring machine 16 is connected to the input side of the control device 20.

As shown in the figure, the work number counter 18 measures the number of waiting works on the processing line between the processing machine 10 and the total measuring machine 16 and waiting for the measurement by the total measuring machine 16. Is. Work counter 1
Reference numeral 8 denotes a first sensor (for example, a limit switch or the like) 46 that detects the carry-out of the work from the processing machine 10 and a second sensor (for example, a limit switch or the like) 48 that detects the carry-in of the work to the total number measuring machine 16. Is connected to and the count value of the number of standby works is incremented by 1 each time the first sensor 46 detects the work carry-out, while the count value is detected each time the second sensor 48 detects the work carry-in. Is subtracted by 1 to measure the current value of the number of waiting works.

The control device 20 includes a CPU, a ROM, and a R.
It is mainly composed of a computer including an AM and a bus, and a fixed point correction routine is stored in advance in its ROM. The control device 20 is also connected to the auxiliary storage device 22, and the total number measuring device 1
It is designed to store all the measured values X input from 6 and the correction values U determined based on them. This is because it is used when an operator diagnoses the machining status after a series of machining is completed.

The main part of the above-mentioned fixed point correction routine is shown in FIG.
A flow chart is shown in FIG. 10, and the configuration of the control device 20 will be described based on these figures, but first, a schematic description will be given.

The controller 20 feeds back the dimension information based on the dimension measured by the total number measuring machine 16 to determine the correction value U of the sizing point for the work to be processed next by the processing machine 10. To do. This processing system is designed to allow a workpiece between the processing machine 10 and the total measuring machine 16 to wait for dimension measurement by the total measuring machine 16. Therefore, the control device 20 assumes a control system in which the correction value U is an input signal, the dimension information is an output signal, and there is a dead time MS between the input signal and the output signal, and the feedback system is used to perform the sizing. Correct the points.
That is, in the present embodiment, the sizing point is one aspect of the "machining condition" in the invention of each claim.

The flow of processing in the control device 20 will be described below.
A simple description is as shown in FIG.

First, as the first step, a total number measuring machine 1
The measurement value X is input from 6, and then, as a second step, a moving average value P is calculated with respect to the measurement values X acquired up to this time in order to remove the adjacent variation from the measurement values X. It In the RAM of the computer, the measured value X
An arithmetic data memory (not shown) for accumulating the above is provided, and the moving average value P is calculated based on the measured value X accumulated in the arithmetic data memory.

Next, as a third step, both-end diameter correction (described in detail later) is performed on the moving average value P,
Further, as a fourth step, the moving average value P and the target value A ₀ of the machining size of the work are calculated based on the moving average value P (also stored in the calculation data memory) whose both end diameters have been corrected. An error value R, which is the difference, and a differential value T and a twice differential value D of the error value R are calculated as the dimension information. Then, as a fifth step, a fuzzy calculation is performed to calculate a correction value U using fuzzy inference based on the dimension information and the number of waiting works measured by the work number counter 18 (corresponding to the dead time MS). Be seen. Then, as a sixth step, the calculated correction value U is corrected by considering its continuity. Thereafter, as a seventh step, it is determined whether or not the corrected correction value U is within the dead zone set in relation to the sizing device 14, and if it is not within the dead zone, the eighth step is performed. The correction value U is the sizing device 14
Sent to.

Further, in the control device 20, not the continuous correction for determining the correction value U of this time each time the workpiece is measured by the total number measuring machine 16, but the intermittent correction for intermittently determining is adopted. There is. Further, as the correction value U is intermittently corrected, the calculation data memory is also cleared intermittently.

It should be noted that the control device 20 is equipped with 7
The measurement value X is individually input for all of the journal surfaces, but basically, it is based on the measurement value X of each of the first journal surface and the seventh journal surface, that is, the measurement value X of each end cylindrical surface. Thus, the correction value U corresponding to each end cylindrical surface in the sizing device 14 is determined.

Although the overall flow of the control device 20 has been briefly described above, each concept in this flow will be described in detail below.

First, calculation of the moving average value P (second in FIG. 11)
Step) will be described. Measured value X is 100
6, which is acquired as time series data, and includes many adjacent variations. Therefore, in the present embodiment, in order to remove the variation between adjacent portions and estimate the true dimension of the work, the weight is applied to the measured value X of this time and the latest measured value X of at least one acquired up to the previous time. A moving average value P with is calculated and used as the true value of the measured value X.

This moving average value P is calculated in principle as follows. That is, based on the latest K (fixed value of 2 or more) measurement values X acquired up to this time, the moving average value of this time is calculated using a calculation formula represented by the following formula (when K = 5). P _i is calculated.

[0046]

[Equation 1]

Here, “i” is the number of works measured by the total number measuring machine 16 (hereinafter referred to as “measured work number”).
Represents

Further, "b _i-4 " to "b _i " are the weighting coefficients of the same number as the number (= K) of the measurement values X necessary for calculating the moving average value P _i of this time.

The value of each of the plurality of weighting factors b is determined in relation to the frequency of the component wave to be removed from the measured value X which is the original variation (that is, the component wave representing the variation between the adjacent portions). However, for example, when the number of standby works existing between the processing machine 10 and the total counting machine 16 is 0, or when it is not 0, it hardly changes, the measured value by the moving average. When the frequency of the component wave to be removed from X hardly changes, the value of each weighting coefficient b can be determined as follows, for example.

First, the angular frequencies of the s component waves to be removed from the original fluctuation are represented by ω ₁ , ω ₂ , ..., ω _j ,.
······ ω _s and create the following formula.

[0051]

[Equation 2]

Then, the coefficient of this equation 1, a _s-1 , ...
·, A _0, ..., respectively 1 to a ₀ of a _s-1, 1, the weighting factor _{b is, b i- (s-} 1), is to determine ..., to b _i.

On the other hand, unlike the above case, when it is not guaranteed that the number of waiting works is always stable, it is possible that the frequency of the component wave to be removed from the measured value X is slightly changed by the moving average. Inevitable. Therefore, in this case, it is desirable to determine the value of each weighting coefficient b as follows, for example. That is, as shown in the graph of FIG. 12, for example, the value of each weighting coefficient b increases almost proportionally as the measured value X multiplied by each weighting coefficient b is newer than the latest measured value X. You just have to decide. By doing so, the component wave with a short period extending over a relatively wide range is removed from the measured value X, and the moving average value P is calculated.

In this case, the value of each weighting coefficient b can be determined by assuming a standard change situation of the number of waiting works and fixing it to an optimum unique size under the assumed change situation. . For example, the above-mentioned b _i-4 , b _i-3 , b _i-2 , b
It is possible to fix _i-1 and b _i to ₁ , 2, 3, 4 and 5, respectively. However, when the value of each weighting coefficient b is fixed in this way, the vibration level of the measured value X (that is, the level of fluctuation when the measured value X periodically fluctuates as the number i of measured works increases). Is high, the moving average value P also vibrates and does not become close enough to the true value, and when the vibration level of the measured value X is low, the moving average value P is sufficiently small with respect to the change of the measured value X. There is a risk of not responding promptly.

Against this background, in the present embodiment, the value of each weighting coefficient b is made variable so that it automatically adapts to the vibration level of the measured value X.

Specifically, the previous value of each weighting coefficient b is used to calculate the provisional value of the moving average value P of this time, and each of the plurality of measured values X used at that time and the moving average value of this time are calculated. The sum of the difference between P and the provisional value is calculated, and the sum is determined as the vibration level of the measured value X (relative vibration level in relation to the previous value of each weighting coefficient b), and the determined vibration level is If it is equal to or smaller than the allowable value A, the provisional value of the moving average value P of this time is determined as the final value. On the other hand, when it is larger than the allowable value A, each weighting coefficient b is increased / decreased by a fixed amount between the highest level and the lowest level, as shown conceptually in the graph of FIG. The value of each weighting coefficient b for which is less than or equal to the allowable value A is determined as the current value of each weighting coefficient b, and the moving average value P calculated using this current value is determined as the final value of this moving average value P. To be done. Therefore, when the vibration level of the measured value X is high, the slope of the straight line representing the distribution of the weighting factor b becomes gentle, and the influence of the current value of the measured value X on the moving average value P is the past value of the measured value X. Is reduced from the effect on the moving average value P, and as a result, the moving average value P is less likely to change significantly depending on the current value of the measured value X. On the other hand, when the vibration level of the measured value X is low, the slope of the straight line representing the distribution of the weighting factor b becomes steep, and the influence of the present value of the measured value X on the moving average value P is The value increases more than the influence on the moving average value P, and as a result, the moving average value P is likely to change significantly depending on the current value of the measured value X.

However, even if the value of each weighting coefficient b is changed over the entire range between the lowest level and the highest level, the value of each weighting coefficient b at which the vibration level becomes the allowable value A or less cannot be obtained. is there. In this case, the value of each weighting coefficient b corresponding to the minimum value of the plurality of vibration levels acquired each time the value of each weighting coefficient b is changed is the weighting coefficient b.
Is determined as the current value, and the moving average value P calculated using this current value is determined as the final value of the current moving average value P.

Although the principle of calculating the moving average value P has been described above, if the principle is adhered to, the moving average value P is moved while the number of measured values X accumulated in the arithmetic data memory does not reach K. The average value P cannot be calculated, and as shown in FIG. 14, neither the error value R nor the differential value T that should be calculated using this cannot be calculated, and as a result, a new correction value U can be calculated. The time that cannot be decided becomes long. In addition, this figure is represented as the number i of measured works increases from the left side to the right side. The same applies to FIGS. 15 and 16 described later.

Therefore, in the present embodiment, when the number of measured values X stored in the arithmetic data memory does not reach K, as an exception, the moving average value is calculated according to a special rule different from the case where it is reached. P is calculated. In addition, this technology is applied for as Japanese Patent Application No. 4-329985 of the present applicant.

The special rules include a substitution type moving average value calculation rule and a variable type moving average value calculation rule. The details will be described below.

First, the substitution type moving average value calculation rule is shown in FIG.
As shown in FIG. 4, each of the K moving average values P scheduled to be present, which does not actually exist, is given a corresponding moving average value P.
Is to be replaced with the measurement value X itself acquired at the same time as the acquisition. This utilizes the property that the measured value X and the moving average value P that are acquired at the same time are originally close to each other, and the calculation of the moving average value P according to this rule means that the replacement type moving average value is calculated. It is called calculation.

In this substitution type moving average value calculation, is the time when the moving average value P is replaced each time a little before the time when the moving average value P is calculated in principle for the first time after that? It is possible to virtually obtain it by replacing the moving average value P with the measured value X, regardless of whether it is long before. However, in this case, there are the following problems. That is, as the ratio of the virtually calculated moving average value P in the L moving average values P used to calculate one differential value T increases, the accuracy of the differential value T decreases. Consequently, there is a problem that the accuracy of the correction value U may decrease.

In order to solve this problem, it is sufficient to limit the number of virtual moving average values P used to calculate one differential value T. The example in the figure is an example in which such a limitation is added, and in this case, only within the past three times from the latest regular moving average value P (that is, the replacement limitation number Z is 3). The calculation of the virtual moving average value P is allowed. In such a case, even if the replacement type moving average value is calculated, the moving average value P should be calculated while the number of measurement values X accumulated in the arithmetic data memory is small. I can't.

On the other hand, the variable moving average value calculation rule prepares a weighted moving average value calculation formula for each of the number of measurement values X (a number smaller than K), and measures stored in the calculation data memory. Select the calculation formula that matches the number of values X,
The rule is to calculate the moving average value P using it (see FIG. 16). Calculating the moving average value P according to this rule is called variable type moving average value calculation. The following weighted moving average calculation formulas can be individually prepared, for example.

[0065]

[Equation 3]

In this example, the moving average value P can be calculated even if the number of measurement values X stored in the arithmetic data memory is one. Therefore, in this example, the moving average value P cannot be calculated because the number of measurement values X accumulated in the arithmetic data memory is small.

In this embodiment, the operator gives an instruction as to whether or not to execute the above-described calculation of the moving average value P by a special method (hereinafter, referred to as "special moving average value calculation"). When the calculation of the special moving average value is instructed, the selection of the type is also made according to the instruction from the operator. That is, when the special moving average value calculation command is issued, either the replacement type moving average value calculation command or the variable type moving average value calculation command is always issued.

Next, the both-end diameter correction (third step in FIG. 11) will be described. In the processing system to which the control device 20 is connected, as described above, the grindstone group 30 is operated based on only the diameters of the two end cylindrical surfaces of all the cylindrical surfaces of the work. Therefore, when the sizing point is corrected without considering the measured values X of the two end cylindrical surfaces but the measured values X of the other cylindrical surfaces, the machining accuracy of each cylindrical surface is the whole. In some cases, it may not be sufficiently uniform.

Therefore, in the present embodiment, the following technique is adopted to solve this problem. That is, as conceptually shown in the graph of FIG. 17, the axial position of each cylindrical surface in the work (“1J” to “7J” in the figure).
And the diameter of each cylindrical surface (that is, the moving average value P)
Assuming that and are in a proportional relationship, a technique called both-end diameter correction is adopted in which the measured values X of the two end cylindrical surfaces are respectively corrected. This technology has been applied as Japanese Patent Application No. 4-329987 of the present applicant.

A specific example of this both-ends diameter correction is as follows. That is, as the both end diameter correction calculation formula,

[0071]

[Equation 4]

The following equation is adopted, and by using this equation, the correction value of the moving average value P of each end cylindrical surface is calculated. However, x: number of journal surface (numbered 1 to 7 from the first journal surface to the seventh journal surface) x ': average value of seven x values y: movement at each value of x Modified value of average value P: Calculated value of moving average value P at each value of x P ': Average value of calculated values of 7 moving average values P

Specifically, for the first journal surface, the corrected value y ₁ of the moving average value P is obtained by substituting 1 for “x” in the above equation, and for the seventh journal surface, , By substituting 7 for “x”,
The corrected value y ₇ of the moving average value P is acquired.

In this embodiment, the operator also gives an instruction as to whether or not to execute the both-end diameter correction.

In the present embodiment, the moving average value P
Both ends are corrected for diameter,
It is also possible to perform both-end diameter correction on the measured value X itself, which is the basis of the moving average value P.

Next, the dimension information acquisition (fourth step in FIG. 11) will be described. As described above, the dimension information acquired for the work includes not only the error value R but also the differential value T and the second differential value D thereof. The error value R is one mode of "dimensional error", the differential value T is one mode of "change tendency of dimensional error", and the twice differential value D is one mode of "change tendency of change tendency".

As described above, the reason why the correction value U is determined based on the parameters other than the error value R is that the differential value T of the correction value U is determined more than when the correction value U is determined based on only the error value R. Alternatively, when the correction value U is also determined based on the twice-differential value D, the actual state of the processing machine 10 can be more accurately estimated, and the correction accuracy of the sizing point is improved. . The technique for determining the correction value U based not only on the error value R but also on the differential value T is disclosed in Japanese Patent Application No.
No. 61305, and a technique for determining the correction value U based on the second derivative D is also filed as Japanese Patent Application No. 4-235402 of the present applicant.

Here, the calculation of the differential value T will be described. As conceptually shown in the graph of FIG. 18, the differential value T is, in principle, L (2 or more or more) including an error value R acquired this time and at least one latest error value R acquired up to the previous time. (Fixed value) error value R is measured work number i
Is assumed to be approximately proportional to the increase of the linear error, the linear regression line to which the L error values R fit is specified, and its gradient is the differential value T (the gradient of the linear regression line is θ radian). ta
(matches n θ). Specifically, as the formula of the primary regression line, for example,

[0079]

[Equation 5]

The following equation is adopted. However, x: value of the number of measured works i x ′: average value of L x values y: true value of error value R at each value of x R: calculated value R of error value R at each value of x ′: Average value of L error value R calculated values, and

[0081]

[Equation 6]

The value of is the differential value T. However, if this principle is followed, similar to the case of calculating the moving average value P, if the number of error values R accumulated in the calculation data memory does not reach L, the differential value T is calculated. I can't.

Therefore, in the present embodiment, in accordance with the calculation of the moving average value P, the formula of the primary regression line is prepared for each of the number of moving average values P (the number smaller than L), A technique is adopted in which an equation that matches the number of error values R stored in the arithmetic data memory is selected and the differential value T is obtained using the equation. This technology is also applied as Japanese Patent Application No. 4-329985.

In this embodiment, the operator also gives an instruction as to whether or not to execute the calculation of the variable differential value.

Next, the calculation of the twice-differential value D will be described. The twice-differential value D is calculated in the same manner as the differential value T.
That is, it is assumed that the latest Q (fixed values of 2 or more) differential values T acquired up to this time are approximately proportional to the increase in the number i of measured workpieces, and these Q differential values T are 1 The secondary regression line is specified, and the gradient thereof is acquired as the second-order differential value D (which matches tan θ when the gradient of the primary regression line is θ radians).

In the present embodiment, the second derivative D
Whether or not to use is also set by the operator.

Further, in the present embodiment, the technique corresponding to the variable differential value calculation related to the differential value T is not adopted for the twice differential value D, but it is of course possible to adopt it.

Next, the fuzzy operation (fifth step in FIG. 11) will be described. Since the fuzzy operation employs techniques such as dead time consideration type correction and fuzzy inference, these techniques will be individually described in detail. explain.

First, the dead time consideration type correction will be described. When the number of waiting works waiting for the measurement by the total number measuring machine 16 changes, it is desirable to change the rule for determining the correction value U according to the number of waiting works. Therefore, in the present embodiment, the number of waiting works is measured as the dead time MS, the determination rule of the correction value U (specifically, a fuzzy rule described later) is changed in accordance with the dead time MS, and the changed determination rule is used. A dead time consideration type correction for determining the correction value U is adopted in accordance with.

In the present embodiment, the operator also gives an instruction as to whether or not to execute this dead time consideration type correction.

Next, fuzzy inference will be described. In this embodiment, there are three types of fuzzy calculation methods. That is, the first method by fuzzy inference using only the error value R and the differential value T as input variables,
The second method by the fuzzy inference using the error value R, the differential value T and the twice differential value D as the input variables, and the third method by the fuzzy inference using the error value R, the differential value T and the dead time MS as the input variables. There is a method of. A specific example of the first method is described in Japanese Patent Application No. 4-
No. 61305, the specific example of the second method is already disclosed in the above-mentioned Japanese Patent Application No. 4-235402, and one specific example of the third method is the Japanese patent application No. 4 of the present applicant. -1
It has already been disclosed in the specification of 58787.

In the present embodiment, the second method is used when the twice differential value use command is issued, and the third method is used when the dead time consideration type correction command is issued. If none of the commands is issued, the first method is selected.

The specific contents of these three types of methods in this embodiment will be described below. However, since these methods have the same basic idea, the first method, that is,
A method of calculating the correction value U by fuzzy inference using only the error value R and the differential value T as input variables will be described as an example.

Data for fuzzy inference is also stored in advance in the ROM of the control device 20. The data for fuzzy inference is (a) inference program,
(b) A plurality of membership functions relating to each of the error value R, the differential value T and the correction value U, (c) A plurality of fuzzy rules etc. that define the mutual relationship between the error value R, the differential value T and the correction value U. is there.

Regarding the error value R, "NB", "NM", "N" as it increases from negative to positive
Seven fuzzy labels are prepared, which sequentially change to "S", "ZO", "PS", "PM" and "PB".
Each membership function is graphically represented in FIG.

Regarding the differential value T, as it increases from negative to positive, "NB", "NS", "Z"
Five fuzzy labels that sequentially change to “O”, “PS”, and “PB” are prepared, and the membership functions of each are shown in a graph in FIG.

As for the correction value U, as it increases from negative to positive, "NB", "NM", "N"
Seven fuzzy labels are prepared, which sequentially change to "S", "ZO", "PS", "PM" and "PB".
Each membership function is graphically represented in FIG. It should be noted that as the correction value U increases, the sizing point increases and the journal portion of the crankshaft increases in diameter, and conversely, when the correction value U decreases, the sizing point decreases and the journal portion decreases in diameter. Will be.

A fuzzy rule group consisting of a plurality of fuzzy rules is originally sufficient for one kind, but in the present embodiment, two kinds are prepared. The reason will be described below.

When the measured value X is stable and the vibration level thereof is low, the correction value U is applied to the error value R and the differential value T.
It is desirable to improve the dimensional accuracy of the work by deciding the correction value U so as to respond sensitively. However, when the contents of the fuzzy rule group are set assuming only such a state, the measured value X becomes unstable due to the vibration of the processing machine 10, etc., and the vibration level thereof becomes high. In this state, when the correction value U is transmitted to the sizing device 14, the vibration level of the measurement value X may increase, and the dimensional accuracy of the work may decrease. Therefore, in the present embodiment, the aggressive fuzzy rule for determining the correction value U so as to be used in a state where the vibration level of the measurement value X is low and to respond sensitively to the error value R and the differential value T is used. A group of
The error value R is used when the vibration level of the measured value X is high.
And the correction value U so as to respond insensitively to the differential value T
And a set of passive fuzzy rules for determining.

Table 1 shows the contents of the aggressive fuzzy rules.
Table 2 shows the contents of the passive fuzzy rules.

[0101]

[Table 1]

[Table 2]

An example of the positive fuzzy rule is If R = NB and T = NS then U = PB as is clear from Table 1, and an example of the passive fuzzy rule is Table 2.
As is clear from, If R = NB and T = NS then U = PS. As is clear from these examples, the positive fuzzy rule group and the negative fuzzy rule group are the same in the fuzzy rule group in which the correction value U is positive even if the error value R and the differential value T are common. It is designed so that it tends to increase more in the passive fuzzy rule group.

In this embodiment, the control device 20
In the initial state of, the aggressive fuzzy rule group is selected from the aggressive fuzzy rule group and the passive fuzzy rule group.

Further, in this embodiment, the aggressive fuzzy rule group is used in the previous fuzzy inference,
Moreover, since the vibration level of the measured value X of this time is high, the passive fuzzy rule group is used in the fuzzy reasoning of this time, and the vibration level of the measured value X of this time is low due to the use of the passive fuzzy rule group. When it becomes, it is designed to use the aggressive fuzzy rules again. When the aggressive fuzzy rule group is used again, the vibration level of the measured value X may immediately become high again. However, if the passive fuzzy rule group is used for a long period of time, the response speed of the correction value U to the change of the true value of the measured value X will be slightly slowed down, and the dimensional accuracy of the work will be deteriorated. , The use period of the passive fuzzy rule group is shortened as much as possible, and the dimensional accuracy of the work is improved.

Furthermore, in the present embodiment, whether or not the vibration level of the measured value X is high is determined when the differential value T is "NB" among a plurality of fuzzy rules in each fuzzy rule group. It is designed to be performed based on the frequency with which the ones corresponding to "PB" (hereinafter, these are referred to as "abnormality fuzzy rules") are matched. Specifically, every time the value of the judgment number counter representing the number of executions of the fuzzy operation becomes larger than the set value B of 2 or more, the number of abnormal times representing the number of times the above-mentioned fuzzy rule for anomalies is met in a plurality of fuzzy operations. It is determined whether or not the value of the counter is larger than the set value C, and when it is large, it is determined that the vibration level of the measured value X is high. Both the judgment number counter and the abnormality number counter are provided in the RAM. Further, the judgment number counter is reset to 0 when the value becomes larger than the set value B. Further, the abnormality number counter is reset to 0 at the same time when the determination number counter is reset.

The reason why two types of fuzzy rule groups are provided and the differences between the fuzzy rule groups have been described above. However, the design concepts of the fuzzy rule groups are basically the same, and the common points are as follows. Will be described.

In each fuzzy rule group, the fuzzy label of the error value R increases (hereinafter, simply referred to as “the error value R increases.” The same applies to other fuzzy variables).
It is designed such that the correction value U decreases as the correction value U increases as the differential value T increases.

This is specifically shown in Table 1 below.
It appears as follows in the fuzzy rule table of. That is, for example, when the differential value T is “NS”,
As the error value R increases, the correction value U becomes "PB", "PB"
M ”,“ PS ”,“ ZO ”,“ ZO ”,“ NS ”and“ NM ”in that order, and when the error value R is“ NM ”, the differential value T is“ NS ”, "ZO" and "P"
As the correction value U increases in the order of “S”, the correction value U becomes “PM”, “P”.
It decreases with "M" and "PS".

The in-process measuring machine 12 may break down for some reason, and in this case, the measurement accuracy of the in-process measuring machine 12 suddenly and greatly decreases, and the dimensional accuracy of the work also suddenly and greatly decreases. Nevertheless, if the correction value U is determined assuming that the in-process measuring machine 12 is normal, the actual dimensional accuracy of the work may deviate from the allowable tolerance range.

In view of such circumstances, the fuzzy rule groups are classified into a case where the measured value X by the total number measuring machine 16 suddenly decreases and becomes considerably small, and a case where the measured value X suddenly increases and becomes considerably large. Each is also designed so that the correction amount U approaches sufficiently 0. In this way, when the in-process measuring instrument 12 fails, the output signal from the in-process measuring instrument 12 is ignored and processing is performed assuming that the sizing points up to the last time are appropriate this time as well. It is possible to maintain high dimensional accuracy of the work without being strongly affected by the failure of.

Specifically, this appears as follows in the fuzzy rule table of Table 1, for example. That is,
When the error value R is "NB" or "NM" and the differential value T is "NB", and when the error value R is "PM" or "PB" and the differential value T is "PB" ”, The correction value U is“ ZO ”.

Next, consideration of continuity (sixth step in FIG. 11) will be described. As described above, the dimensional error of the work generally increases almost in proportion to the increase in the number of measured works i. Therefore, the correction value U of the sizing point should be continuous, that is, It is desirable to change smoothly as the processing progresses in order to suppress dimensional variation of the work.

Therefore, in this embodiment, paying attention to the fact, as shown conceptually by the graph in FIG.
Ignoring the continuity is determined correction value U is, it provisional value (hereinafter, referred to. In addition, "provisional correction value U" is different from the provisional correction value U _P below) are and have been obtained until the current It is assumed that the latest M (fixed values of 2 or more) provisional correction values U are approximately proportional to the increase in the number of measured workpieces i, and the first-order regression similar to the above case is performed for these M provisional correction values U. The line formula is specified. Then, the true value of the current correction value U using equation is estimated, it is the final value of the correction value U (hereinafter, referred to as "final correction value U ^*". Note that the final correction value U _F below Is different). In addition, this technology is
The present application is filed as Japanese Patent Application No. 4-61306.

Specifically, as an equation of the primary regression line, for example,

[0115]

[Equation 7]

The following equation is adopted. However, x: value of the number i of measured works x ′: average value of M x values y: true value of provisional correction value U at each value of x U: calculation of provisional correction value U at each value of x Value U ′: average of calculated values of M provisional correction values U

Then, by substituting the value of the number i of measured works for this time into "x" in the above equation, the final correction value U ^{* for} this time is obtained.

In this embodiment, the operator also gives an instruction as to whether or not to execute the continuity-aware correction.

Further, when the operator issues the continuity-aware correction command (however, the above-mentioned second derivative value use command is not issued), the final correction value U ^* is acquired from the measured value X. The process up to this point is representatively shown conceptually in FIG. This figure is represented as the value of the measured work number i increases from the left side to the right side. As is clear from the figure, when the accumulation of the measured value X in the calculation data memory is started from the non-accumulation state, (K
+ L + M-2) one final correction value U ^* will be obtained only when the measurement values X are accumulated.
This is one mode of "a plurality of set values" relating to the accumulation of the measured value X, which will be described later.

Next, the dead zone consideration (seventh step in FIG. 11) will be described. For the transmission of the final correction value U ^* , the dead zone in relation to the sizing device 14 is set,
If the final correction value U ^* determined each time is within the dead zone, the transmission of the final correction value U ^{* to} the sizing device 14 is omitted. This state is conceptually shown in the graph of FIG. Further, in this embodiment, a range having a width of N around 0 is defined as one mode of the "dead zone". The technique of setting a dead zone in the correction value U has been filed as Japanese Patent Application No. 4-278146 of the present applicant.

Next, the intermittent correction will be described in detail. At the time of sizing point correction, a correction method that is a continuous correction that determines the correction value U of the sizing point of the work to be processed next by the processing machine 10 every time the dimension of the work is measured by the total number measuring machine 16. Can be adopted. However, when this continuous correction is adopted, there are the following problems.
That is, since the correction value U has to be individually determined for all the works measured by the total number measuring machine 16, there is a problem that the control device 20 is heavily burdened.

In order to solve this problem, in this embodiment, a correction method called intermittent correction is adopted.

This intermittent correction is conceptually shown in the graph of FIG. This graph is acquired when there are a plurality of standby works between the processing machine 10 and the total measuring machine 16, and the “measurement delay” in this graph means the processing machine 10 and the total measuring machine 16. Equivalent to the number of waiting works between and. In addition, “U _i ” is the correction value of this time and is “U _{i + 1} ”.
Indicates the next correction value. Therefore, the influence of the correction value U _i of this time appears on the dimension error only after the measurement delay, and similarly, the influence of the next correction value U _{i + 1} also appears on the dimension error after the measurement delay. In addition, this graph is based on the assumption that, when a plurality of works are machined in order under the same fixed size point, the dimensional error of each work increases almost in proportion to the increase in the number of measured works i. It is also acquired by. Note that, for example, the wear of the grindstone as the processing tool of the processing machine 10 is considered as the cause of the increase in the dimensional error. The above situation is the same in the following graphs.

The present applicant has devised two modes as a method for performing this intermittent correction. Hereinafter, each of these methods will be described in detail.

First Method of Intermittent Correction As described above, this fixed-point correction apparatus is used with a processing system that allows a workpiece to exist between the processing machine 10 and the total counting machine 16. Since it should be, it is not always the case that the workpiece machined under the sizing point affected by the previous correction value U is measured by the total number measuring machine 16 immediately after that, and the measurement of several different workpieces is performed. In some cases, it may be measured for the first time after passing. Therefore, when it is necessary to directly reflect the influence of the previous correction value U on the present correction value U, at least 1 processed under the sizing point affected by the previous correction value U is processed. One workpiece is a total measuring machine 1
It is desirable to determine the correction value U of this time each time it is measured by 6.

Against this background, in the first method, as shown conceptually in the graph of FIG.
When the number of becomes set plurality or, to determine the current correction value U _i based on their accumulated latest settings plurality of measurement X, affected by the current correction value U _i After the time when the head correction target work to be measured first by the total number measuring machine 16 out of at least one work processed under the sizing point ends the measurement (for example,
Immediately after the end of the measurement), the accumulation of the measured value X is resumed from the non-accumulation state.

In this system, for example, after the correction value U _i for this time is determined and transmitted to the sizing device 14, the correction value U for the next time is calculated.
_In the correction interval period, which is an intermediate period until _{i + 1} is determined and transmitted to the sizing device 14, the correction value U is not determined, and the sizing point in the sizing device 14 is maintained at the same value. It can be implemented as an aspect. Then, in this case usually under the premise of a proportional relationship between the measured workpiece number i and dimensional error is established, the magnitude of the present correction value U _i is the effect of the current correction value U _i It is determined that the dimensional errors of the plurality of workpieces that have been subjected to the exposure approach zero as a whole.

However, this embodiment has the following problems. That is, there is a problem that the execution timing of each correction is uniquely determined by the number of accumulated measurement values X regardless of the actual fluctuation of the measurement value X, and each correction is not executed at the time when it is really necessary. It will happen.

To solve this problem, as described above, when the correction value U is transmitted, the dead zone is set in relation to the sizing device 14, and when the determined correction value U is within the dead zone. Does not transmit the correction value U to the sizing device 14, does not clear the calculation data memory, waits for acquisition of a new measurement value X, and then decides the correction value U again. By doing so, each correction can be performed in a timely manner at the time when it is really necessary.

However, in this way, if an unexpected change occurs in the measurement value X after the end of each correction, it is not possible to quickly respond to the change and correct the sizing point. Can not. When an unplanned change occurs in the measured value X after the end of each correction, the unplanned change is accumulated in the calculation data memory and reflected in the next correction value U _{i + 1} . In this way, the sizing point cannot be corrected in response to the unexpected change without waiting until the next correction. Therefore, there is a problem that the dimensional error of the work does not approach 0 sufficiently when the measured value X changes unexpectedly after the completion of each correction.

In order to solve this problem, the first method may be implemented in the following manner. That is, FIG.
As shown conceptually in the graph in FIG. 7, one intermittent correction is followed by a main correction which is an intermittent correction in the above-described mode (for example, determining “U _i ” in FIG. 26) and an auxiliary correction. By performing the above, the sizing point may be corrected in such a manner that an unplanned change in the measured value X after the main correction is completed is quickly corrected by the auxiliary correction.

Here, the "main correction" means that the measured values X by the total number measuring device 16 are successively accumulated, and when the number of accumulated measured values X is equal to or more than a set number, the latest accumulated values are stored. This provisional correction value U based on a plurality of set measurement values X
_P is determined and used as the final correction value U _F.

The "auxiliary correction" means that the measurement value X is continuously accumulated even after the main correction is completed, and the main correction is performed after the main correction is completed (for example, immediately after the main correction is completed). At least 1 processed under sizing point affected by
Before the time when the measurement is finished (for example, until the time when the measurement is finished) for the workpiece that is processed only once before the head correction target work that is to be first measured by the total number measuring machine 16 among the individual works. ), each time the workpiece is measured by the total number measuring instrument 16, based on the stored latest settings plurality of measurement values X, to determine each time the provisional correction value U _P according to the same rules as in the main correction, the It determined each time the provisional correction value U preceding temporary correction value from the _P U _P
The value obtained by subtracting is determined as the final correction value U _F for each time.

[0134] The difference from the in assisting correction is not transmitted is as in the main correction is determined according to the same rule correction value U provisional correction value U _P is directly to the sizing device 14, the previous provisional correction value U _P However, the reason for this will be described below.

In the auxiliary correction, the correction value U should normally be determined according to the same rule as in the main correction, based on the measured value X of the work which has been affected by the main correction performed prior to the correction. . However, the work affected by the main correction is not always measured by the total number measuring machine 16 immediately after machining, and may be measured only after some other work is measured. Therefore, in the present embodiment, the influence of the main correction is duplicated and is not reflected once at the head correction target work related to the main correction so as not to be reflected in the sizing point corresponding to the work to be processed next. only the previously machined workpiece until such time prior to end measurement, correction values U determined according to the same rules as in the main correction based on each time of measurement X is a provisional correction value U _P,
The final correction value U is obtained by removing the effect of the main correction.
It is said to be _F. Although the relationship between the main correction and the auxiliary correction has been described above, the same applies to the relationship between one time and the next time in the auxiliary correction.

In the mode in which the main correction and the auxiliary correction are performed, the auxiliary correction can be always executed until the end time of the intermittent correction to which the main correction belongs, that is, immediately before the start of the next main correction. However, in this case, there is a problem that the control device 20 itself is slightly burdened.

In order to solve this problem, it is sufficient to limit the number of executions of auxiliary correction. That is, the number of determinations of the final correction value U _F in the series of auxiliary corrections is measured, and when the measured number of determinations reaches the set value, the series of auxiliary corrections is ended. However, with this measure,
The end time of the auxiliary correction is fixed in relation to the end time of the main correction, and the execution time of the auxiliary correction is optimal for responding to an unexpected change in the measured value X after the end of the main correction. There is a problem that it is not limited.

In order to solve this problem, auxiliary correction may be carried out in the following manner. That is, when the number of determinations of the final correction value U _F in the series of auxiliary corrections reaches the set value, a plurality of final correction values U determined in at least the series of auxiliary corrections of the main correction and the series of auxiliary corrections. _If the sum of _F is not substantially 0, the series of auxiliary corrections is ended, but if it is substantially 0, it is estimated that at least the current execution time of the auxiliary correction was not appropriate. The auxiliary correction of this time may be continued and the measurement of the number of determinations of the final correction value U _F may be restarted from 0.

In the present embodiment, as a method of determining the correction value, either the main correction only without the auxiliary correction or the main correction as well as the auxiliary correction is instructed by the operator. It is selected according to. That is, if the auxiliary correction command is issued, the latter method is selected, and if it is not issued, the former method is selected.

Further, in the present embodiment, as a method of the auxiliary correction, either the method of continuing the auxiliary correction or the method of not performing the auxiliary correction can be selected according to the operator's instruction. There is.

Furthermore, in the present embodiment, as a method for continuing the auxiliary correction, the provisional correction value U _P is used as it is as the final correction value U _{F in} the first auxiliary correction to be continued.
As a method to continue assisting correction (hereinafter, referred to as "auxiliary compensation resumption method") and, on the difference of the final correction value U _F from the previous value of the current value of provisional correction value U _P from initial to be continued auxiliary correction One of the method of determining and continuing the auxiliary correction (hereinafter, referred to as "auxiliary correction extension method") is also selected according to the operator's instruction. The command for selecting the former method is called the "auxiliary correction restart command", and the command for selecting the latter method is called the "auxiliary correction extension command". If neither of these commands is issued, ,
It is determined that the auxiliary correction continuation permission command has not been issued.

Second Method of Intermittent Correction When performing the intermittent correction by the above-mentioned first method, when the standby work exists between the processing machine 10 and the total number measuring machine 16, the current correction The accumulation of the measured value X cannot be started immediately after the determination of the value U. Therefore, the time (hereinafter, referred to as “correction interval time”) required from the determination time of the current correction value U to the determination time of the next correction value U is as shown in FIG.
As shown in, the measurement delay time in which the head correction target work related to the correction value U of this time reaches the total measuring machine 16 (corresponding to the number of standby works existing between the processing machine 10 and the total measuring machine 16) ), And thereafter, the accumulation of the measured values X is started, and the required number of accumulated times until a plurality of set measured values X are accumulated is the sum. Therefore, when it is unavoidable that many standby works exist between the processing machine 10 and the total number measuring machine 16, it is unavoidable that the correction interval time becomes long.

This second method was devised to solve this problem, and as shown conceptually in the graph of FIG. 28, the measured values X by the total number measuring machine 16 are successively accumulated, When the number of accumulated measured values X exceeds the set number, the latest accumulated plural measured values X are set.
The correction value U of this time is determined based on the above, and the accumulation of the measurement value X is restarted from the non-accumulation state after the determination time of the correction value U of this time (for example, immediately after the determination time of the correction value U of this time). ,
Intermediate from the restarting time to near the time when the measurement ends for the work processed only once before the top correction target work related to the current correction value U (just before, slightly before, or slightly after that) The period is determined based on the actual measurement value X of each time and the correction value U of this time every time the work is measured by the 100% measuring machine, and each work is influenced by the correction value U of this time. Predicting the values measured for each of these workpieces, assuming that they have been machined under a certain point, and using the predicted measurement values X as the actual measurement values X
It is regarded as and accumulates.

In the present embodiment, as an example of the prediction, the actual measured value X is added to the actual measured value X in the intermediate period to obtain the actual measured value X. A data shift process that shifts only the data is used.

Also in the second method, as in the case of the first method, one intermittent correction includes a main correction and an auxiliary correction which can be continued with a limited number of times (this is shown in FIG. 29). (It is conceptually shown in the graph), and the concept of the dead zone of the correction value U is adopted, and it is implemented as a mode in which the continuation method of the auxiliary correction can be selected. Further, in the present embodiment, it is also possible to select between the first method and the second method in accordance with the operator's instruction. Specifically, the operator gives an instruction as to whether or not to permit the data shift process, and if the data shift process is permitted, the data shift process is selected, and if not, the first method is selected.

It should be noted that the second method is the measurement value prediction technique, that is, the present method out of at least one workpiece machined under the sizing point affected by the previous correction value U. After the correction value U is determined, for each of the ones measured by the total number measuring machine 16, each work is affected by the correction value U of this time based on the actual measurement value X of each time and the correction value U of this time. It was acquired by applying the technique of predicting the value to be measured for each of these workpieces, assuming that they were machined under the sizing point that was subjected to The value prediction technique can also be applied to the continuous correction.

When a standby work exists between the processing machine 10 and the total measuring machine 16, even when performing continuous correction, the work affected by the previous correction value U is immediately measured by the total measuring machine. The situation is the same that 16 cannot measure. Therefore, in this case, the correction value U of this time is determined by a statistical method based on the experimental result, the simulation result, and the like. Then, this measurement value prediction technique can be used instead of or together with the statistical method.

Next, the relationship between the automatic correction by the control device 20 and the manual correction by the sizing device 14 will be described.

When the sizing device 14 performs the manual correction, the control device 20 prioritizes the manual correction over the automatic correction and ensures the accuracy of the automatic correction even immediately after the manual correction. Is designed to. In particular,
The control device 20 sequentially monitors whether or not the manual correction is performed in the sizing device 14, and if the manual correction is not performed, the automatic correction is performed. If the manual correction is performed, the automatic correction is interrupted and the manual correction is performed. Of the measured value X after the time when the head correction target work processed under the sizing point affected by is measured by the total number measuring machine 16 (for example, immediately after the head correction target work is measured). It is also designed to restart the automatic correction by restarting the storage in the calculation data memory from the non-storage state. The controller 20 is designed to also use the past measured value X to determine the automatic correction value. Moreover, in the processing system in which the controller 20 is used, the processing machine 10 and the total measuring machine are used. Since the existence of a standby work between 16 and 16 is allowed, the measurement value X of the work not affected by the manual correction is accumulated without distinguishing it from the measurement value X of the affected work, If the automatic correction value is determined based on the accumulated measurement value X, the accuracy of the automatic correction value may decrease due to the measurement value X of the work that is not affected by the manual correction. is there.

Further, in this embodiment, the sizing device 14
The monitoring of the presence or absence of the manual correction is performed at the start of the control device 20 and prior to each time the control device 20 tries to transmit the automatic correction value each time. The presence or absence of manual correction in the sizing device 14 is monitored even when the control device 20 is started, because the sizing device 14 may be manually corrected while the control device 20 is stopped.

The outline of the sizing point correction by the control device 20 has been described above. The details will be described below with reference to the flowcharts of FIGS.

First, step S1 in FIG. 5 (hereinafter simply referred to as "S1". The same applies to other steps).
At, the numerical values and commands are input as parameters from the auxiliary storage device 22. Here, the “numerical value” means the above-mentioned initial value of the weighting coefficient b relating to the moving average value P, the replacement limit number Z,
The auxiliary correction limit number S and the like are meant, and the "command" means the above-mentioned special moving average value calculation command and the like.

Subsequently, in S2, it is determined whether or not the sizing device 14 has a function of transmitting the latest manual correction value and the sizing point to the control device 20 (hereinafter referred to as "manual correction value transmission function"). Is determined. Here, the case of having a manual correction value transmission function means that the sizing device 14 sets the manual correction value input to the sizing device 14 by the operator through the keyboard 50 and the sizing point reflecting the manual correction value. Not only it is stored in the RAM, but it is designed to be transmitted to the control device 20 spontaneously. On the other hand, the case without the manual correction value transmission function means that the sizing device 14 has been input. This is a case in which the manual correction value and the sizing point reflecting the manual correction value are only stored in the RAM of the device itself and are not designed to be transmitted to the control device 20 spontaneously.

Since the sizing device 14 has the manual correction value transmitting function as described above, the determination in S2 is YE.
S (represented by “Y” in the figure. The same applies to other steps). Therefore, the process proceeds to S3, in which the control device 20 enters a reception permission state that permits reception of a signal representing the latest sizing point from the sizing device 14, and the latest sizing point is received. The latest received sizing point is stored in the RAM of the control device 20 and also stored in the auxiliary storage device 22.

Then, in S4, R of the sizing device 14
From the state of the AM flag, it is determined whether or not there is manual correction. Assuming that there was no manual correction, the determination is NO.
Then, the process immediately proceeds to S8, but if it is assumed that the determination is YES, then in S5, the control device 20
However, the reception permitting state for permitting the reception of the signal indicating the latest manual correction value from the sizing device 14 is received, and the latest manual correction value is received. The latest received manual correction value is stored in the RAM of the control device 20, and is further stored in the auxiliary storage device 2
Stored in 2. Then, in S6, the operation data memory is cleared. With the manual correction, all the data accumulated in the calculation data memory is erased.
Then, it transfers to S8.

The case where the determination in S2 is YES has been described above. However, assuming that the sizing device 14 does not have the manual correction value transmitting function, the determination is NO (represented by "N" in the figure). The same applies to other steps).
Then, in S7, the control device 20 causes the sizing device 14 to
Then, the latest sizing point is read from the RAM and stored in the RAM of the control device 20, and the latest sizing point is stored in the auxiliary storage device 22.

Now, the purpose of the control device 20 to monitor the latest manual correction value and the sizing point in the sizing device 14 will be described.

First, the purpose of the control device 20 to monitor the latest sizing point will be described. The controller 20 corrects the sizing point of the sizing device 14 (the amount by which the current sizing point is changed).
Is automatically determined, and the sizing device 14 corrects its own sizing point according to the determined correction value. But,
There is a limit to the range that the sizing point can take, and when the sizing point beyond that is determined, the operation of the sizing device 14 is stopped. Therefore, in the present embodiment, the sizing device 1
The latest sizing point 4 is sequentially monitored, and if the sizing point is corrected by the automatically determined correction value and the allowable range is exceeded, the sizing device 14 for the automatic correction value is used. Sending to is prohibited. As described above, in order to prevent the sizing point of the sizing device 14 from exceeding the allowable range by one-sided automatic correction that does not consider the circumstances of the sizing device 14, the control device 20 controls the latest sizing device 14. The sizing point is monitored. The process of prohibiting the transmission of the automatic correction value when the sizing point exceeds the allowable range in the sizing device 14 is realized by executing another routine (not shown).

Next, the purpose of the controller 20 to monitor the latest manual correction value will be described. As described above, when the operator issues a data shift processing command that permits the use of the data shift processing, the control device 20 determines the processing machine 1 based on the latest measured value X by the total counting machine 16.
Predicting the values measured for each workpiece, assuming that each workpiece that has been machined by 0 but not yet measured by the probing machine 16 has been machined under the latest sizing point. At this time, when the latest sizing point is automatically corrected by the control device 20, the above-described prediction is performed by adding the latest automatic correction value to the latest measurement value X by the total number measuring machine 16, On the other hand, if the sizing device 14 itself has manually corrected it, the latest measured value X
The above prediction is performed by adding the latest manual correction value to. As described above, in order for the control device 20 to predict the measured value X in consideration of the influence of the manual correction value in the sizing device 14, it is necessary to monitor the latest manual correction value in the sizing device 14. .

Whether the sizing device 14 has or does not have the manual correction value transmission function, the measured value X measured by the total number measuring device 16 in step S8 is not the total number measuring device 16 yet. It is determined whether there is any information that has not been transmitted from the control device 20 to the control device 20. This time, assuming that there is no such measured value X, the determination becomes NO, and the process proceeds to S9.

At S9, similarly to S2, it is determined whether the sizing device 14 has a manual correction value transmitting function. Since the sizing device 14 has a manual correction value transmission function, the determination is YES, and in S10, it is determined whether or not there is a manual correction in the sizing device 14 as in S4.

Assuming that there was no manual correction this time,
The determination becomes NO, and in S11, it is determined whether or not the keyboard (not shown) connected to the control device 20 has been operated by the operator, that is, the presence or absence of key input by the operator. If not, the determination is NO and the process immediately returns to S8. If there is, the determination is YES, data is input from the keyboard in S12, and the parameter is changed according to the data in S13. The changed parameters are stored in the auxiliary storage device 22, after that, in S14, the calculation data memory is cleared, and then the process returns to S8.

On the other hand, assuming that there is manual correction this time, the determination in S10 is YES, and in S15, the latest manual correction value is received and stored from the sizing device 14 in the same manner as in S5. Then, in S16, a work waiting flag described later is turned on, and in S17, the operation data memory is cleared. Then, the process returns to S8.

On the other hand, assuming that the sizing device 14 does not have the manual correction value transmitting function, the determination in S9 is N.
When it becomes O, the latest sizing point is read from the sizing device 14 in S18 and is stored in the RAM and is stored in the auxiliary storage device 22. Then, in S19, the last sizing point is read from the auxiliary storage device 22. The size is entered. Then, in S20, it is determined whether or not the current sizing point has been changed from the previous sizing point. In other words, whether or not the manual correction is performed in the sizing device that does not have the manual correction value transmission function is determined from the change state of the sizing point. Assuming there is no change in the sizing point this time,
The determination is NO, and the process immediately proceeds to S11. However, assuming that the sizing point has been changed, the determination of S20 is YES, and the work waiting flag is turned on in S21.
In S22, the operation data memory is cleared, and then the process proceeds to S11.

The case has been explained above where there is no measurement value X to be transmitted by the total number measuring instrument 16, but if there is, the determination at S8 becomes YES, and at S23, the measurement value X is determined from the total number measuring instrument 16. Is entered. Measured value X
Are individually entered for all seven journal surfaces. The measured value X is accumulated in the operation data memory and is also stored in the auxiliary storage device 22, and thereafter, S2 of FIG.
Go to 4.

At S24, it is determined whether or not the operator has issued a data shift processing command based on the values of the parameters. Hereinafter, first, a case where the data shift processing command is not issued will be described.

In this case, the determination in S24 is NO and S24
At 25, it is determined whether or not the work waiting flag is ON.

This work wait flag is the sizing point in the sizing device 14 and is the first one of at least one work machined under the influence of the latest manual correction value or automatic correction value. Is for monitoring whether the top correction target work is measured by the total number measuring machine 16 or is waiting for the measurement. When the work waiting flag is OFF, it indicates that the head correction target work has finished the measurement, that is, it is not in the work waiting state. On the other hand, when it is ON, the head correction target work has not finished the measurement, that is, the work waiting state. Indicates that there is. This work waiting flag is provided in the RAM,
It is turned on when the power of the computer is turned on, and by the execution of another program (not shown), the OF correction is performed each time the workpiece to be head-corrected is finished by the total number measuring machine 16.
F will be done. Further, by executing this routine, it is turned on each time manual correction is performed and each time intermittent correction is completed. Assuming that the work waiting flag is not ON this time, the determination in S25 is NO and S26
Move to.

In S26, the past measured value X is input from the arithmetic data memory. Then, in S27, it is determined whether or not the current moving average value P can be calculated.
It is determined whether or not the number of measured values X stored in the arithmetic data memory is K or more. This time, assuming that the number of accumulated measurement values X is not K or more, the determination is NO, and in S28, it is determined whether or not there is a special moving average value calculation command. If not, the determination is NO, and the process immediately returns to S8. Therefore, in this execution of this routine, the automatic correction value is eventually set to zero.

On the other hand, if there is a special moving average value calculation command, the determination in S28 is YES, and in S29, the presence or absence of the variable moving average value calculation command is determined. If not, the determination is no and the process moves to S30. Since the variable type moving average value calculation command and the replacement type moving average value calculation command are alternative commands, if there is no variable type moving average value calculation command, there is always a replacement type moving average value calculation command. Become.

In S30, it is determined whether the replacement type moving average value can be calculated. Specifically, whether the number of measurement values X stored in the operation data memory is smaller than K (the number of measurement values X required to calculate the moving average value P in principle) -Z (replacement limit number). If not, and if so,
It is determined that the substitution type moving average value calculation is impossible (correctly, prohibited), and it is determined otherwise (correctly, permitted). If it is not possible, the process returns to S8, but if it is possible while the execution of this routine (execution of the steps after S8) is repeated many times, the determination becomes YES, and the measured value X of this time is kept as it is in S31. The moving average value P of this time is set, and it is stored in the operation data memory in S32, and
It is stored in the auxiliary storage device 22. Then, it transfers to S37.

On the other hand, if there is a variable moving average value calculation command, the determination in S29 becomes YES, and in S33, the moving average value P is calculated by the variable moving average value calculation method.
Is calculated and stored in the auxiliary storage device 22 while being stored in the calculation data memory in S34. Then, it transfers to S37.

After that, assuming that the number of measured values X accumulated in the arithmetic data memory becomes K or more during the execution of this routine many times, the determination in S27 becomes YES and S35. In, the moving average value P is calculated in principle.

Details of S35 will be described below with reference to the flowchart of FIG.

First, in S201, the value of each weighting coefficient b is read from the RAM, and then in S202,
The moving average value P is calculated by using the above-mentioned weighted moving average value calculation formula based on the plurality of measured values X accumulated in the calculation data memory under the read value of each weighting coefficient b. It

Then, in step S203, the sum (absolute value) of the difference between each of the plurality of accumulated measurement values X (used for calculating the moving average value P) and the calculated moving average value P is calculated. Is determined by the vibration level of the measured value X. In this step, the value of each weighting coefficient b, the moving average value P, and the vibration level are further stored in the RAM in association with each other. Then, in S204, it is determined whether or not the determined vibration level is higher than the allowable value A. Assuming that this time is not large, the determination is NO, and in S205, the moving average value P calculated by the current execution of S202 is determined as the current moving average value P, and thereafter, in S206, each of the weights described above is determined. The value of the coefficient b is stored in the RAM as the current value of each weighting coefficient b in preparation for the next calculation of the moving average value P. This is the end of the one-time execution of S35.

On the other hand, assuming that the vibration level determined in S203 is larger than the allowable value A, S2
The determination of 04 is YES, and in S207, the value of each weighting coefficient b is changed according to a certain rule. afterwards,
In S208, whether or not the change is made over the entire range between the highest level and the lowest level of each weighting factor b,
That is, it is determined whether the change is completed. This time, assuming that it has not finished, the determination is NO,
It returns to S202.

After that, in the steps after S202, the moving average value P is calculated based on the plurality of measurement values X same as the previous time under the value of each weighting coefficient b different from the previous time, and the vibration of this time is calculated. The level is also calculated. S202
The execution of steps 208 to 208 ends when the value of the weighting coefficient b at which the vibration level is equal to or lower than the allowable value A is acquired and the determination in S204 is NO, and when the vibration level does not become equal to or lower than the allowable value A. Ends, the determination in S208 is YE
There are cases where the process ends when S is reached. In the former case, the steps from S205 onward are executed as in the case described above.

On the other hand, in the latter case, that is, when the execution of S202 to S208 is completed by the determination of S208 being YES, in S209, a plurality of vibration levels stored in the RAM (several times of S202 to 208 are stored a plurality of times). (The one obtained by executing the above) is searched, and the moving average value P corresponding to the minimum vibration level is R
It is read from AM, and it is determined as the moving average value P of this time. Then, in S206, the moving average value P
The value of each weighting coefficient b corresponding to is read from RAM,
The present value of each weighting coefficient b is stored in the RAM in preparation for the next calculation of the moving average value P.

When S35 is executed as described above, the moving average value P of this time determined in S36 of FIG. 6 is accumulated in the arithmetic data memory and also saved in the auxiliary storage device 22. Then, it transfers to S37.

In S37, the presence or absence of the both-ends diameter correction command is determined. If not, the determination is NO, and the process immediately proceeds to S39 in FIG. 7, but if there is, the determination is YES, and in S38, the two The both-end diameter correction is performed on the moving average value P of the end cylindrical surface, and the content of the calculation data memory is changed according to the result. Then, it transfers to S39 of FIG.

In this S39, the value obtained by subtracting the target value A ₀ of the work size from the current moving average value P is set as the current error value R, and subsequently in S40, it is stored in the calculation data memory. Auxiliary storage device 22
Stored in.

Thereafter, in S41, it is determined whether or not the differential value T can be calculated. It is determined whether or not the number of moving average values P stored in the arithmetic data memory is L or more. This time, assuming that the number of moving average values P is insufficient, the determination becomes NO, and the process proceeds to S42.
In S42, the presence / absence of the variable differential value calculation command is determined. If not, the determination is NO, the process immediately returns to S8, and the current execution of this routine ends, but if there is, the determination is YES and S43. At, it is determined whether or not the moving average value P stored in the arithmetic data memory is two or more, that is, whether or not the variable-type differential value calculation is possible. If not, the determination is NO. And immediately returns to S8, but if possible, the determination is YES, and S
In 44, the differential value T of this time is calculated by the variable differential value calculating method, and in S45, the differential value T is accumulated in the arithmetic data memory and is saved in the auxiliary storage device 22. Then, the process proceeds to S48.

After that, assuming that the number of moving average values P accumulated in the arithmetic data memory becomes L or more during the execution of this routine many times, S41
Is YES, the differential value T is calculated in principle in S46, and the differential value T is accumulated in the operation data memory and saved in the auxiliary storage device 22 in S47. Then, the process proceeds to S48.

In S48, the presence / absence of the twice-differential-value use command is determined, and if there is, the determination is YES and S4
In 9, it is determined whether the differential value D can be calculated twice.
It is determined whether or not the number of differential values T accumulated in the arithmetic data memory is Q or more. This time, assuming that the number of accumulated differential values T is not equal to or more than Q, the determination becomes NO, the process immediately returns to S8, and the current execution of this routine ends. If it is assumed that the number of differential values T accumulated in the arithmetic data memory becomes Q or more during the execution of this routine many times, S4
The determination result in 9 is YES, the differential value D is calculated twice as described above in S50, and is stored in the auxiliary storage device 22 while being accumulated in the arithmetic data memory in S51. Then, it transfers to S55.

On the other hand, if there is no second differential value use command, the determination in S48 is NO, and in S52 it is determined whether there is a dead time consideration type correction command. If NO, the determination is NO and the process immediately proceeds to S55, but if YES, the determination is YES, and the dead time MS is input from the work number counter 18 in S53 and is stored in the arithmetic data memory in S54. , Are stored in the auxiliary storage device 22. Then, it transfers to S55.

Details of S55 will be described with reference to the flowchart of FIG. First, in S301, RA
The abnormality flag is read from M. This abnormal flag is
When it is provided in the RAM, OFF indicates that the vibration level of the measured value X is low, and ON indicates that the vibration level is high. It is turned off when the computer is turned on,
After that, it is changed by executing S65 or 70 described later. Next, in S302, the abnormality flag is set to O.
It is determined whether or not N is set. Assuming that it is not turned on this time, the determination is NO, and in S303, the provisional correction value U is calculated by the fuzzy calculation using the aggressive fuzzy rule group (represented by "aggressive rule" in the figure). It This is the end of the one-time execution of S55.

On the other hand, this time, the abnormality flag is ON.
If YES, the determination in S302 is YES, and in S304, the provisional correction value U is calculated by a fuzzy calculation using a negative fuzzy rule group (represented by “negative rule” in the figure). This is S55
One execution of is completed.

In this embodiment, as soon as the abnormality flag is turned from OFF to ON, the fuzzy rule group to be used is changed from the positive one to the passive one, and conversely, when it is turned from ON to OFF. , It was supposed to be changed from the passive one to the positive one, but it is possible to give the hysteresis characteristic to the change of the fuzzy rule group. That is, for example, the abnormality flag changes from OFF to ON
Immediately after that, the fuzzy rule group to be used is not changed from the positive one to the negative one, and is changed only after the abnormal flag is ON for the set number of times. It is possible. By doing so, it is possible to satisfactorily avoid the situation where the measured value X hunts and becomes unstable due to frequent changes in the fuzzy rule group.

When S55 is executed as described above, the calculated provisional correction value U is stored in the operation data memory and stored in the auxiliary storage device 22 in S56 of FIG. Then, it transfers to S57 of FIG.

In S57, the presence / absence of the continuity-aware correction command is determined. If not, the determination is NO, and S
At 58, the provisional provisional value U is unchanged and the final correction value U ^*
Then, in S59, it is stored in the auxiliary storage device 22. On the other hand, if there is the continuity consideration type correction command, the determination in S57 is YES, and in S60, it is determined whether or not the continuity consideration type correction is possible. It is determined whether or not the number of provisional correction values U stored in the calculation data memory is M or more. If it is assumed that the number of accumulated provisional correction values U is not M or more this time, the determination is NO, the process immediately returns to S8, and the current execution of this routine ends. After that, assuming that the number of provisional correction values U accumulated in the operation data memory becomes M or more during the execution of this routine many times, S
The determination of 60 is YES, and in S61, the final correction value U ^* is calculated as described above, based on the latest M provisional correction values U stored in the calculation data memory. After that, in S62, it is stored in the operation data memory and is stored in the auxiliary storage device 22.

When the execution of S59 or S62 is completed, the presence or absence of the auxiliary correction command is determined in S63 of FIG. Assuming that this time does not exist, the determination is NO, and S64
At, it is determined whether or not the final correction value U ^* at this time should be transmitted to the sizing device 14, that is, whether or not the final correction value U ^* is out of the dead zone. If it is assumed that it is within the dead zone this time, the determination is NO, and in step S65, the fuzzy rule matched in the fuzzy operation is stored in the auxiliary storage device 22. Then immediately S8
Then, the current execution of this routine ends.

On the other hand, assuming that the final correction value U ^* is out of the dead zone, the determination in S64 is YES, and in S66, it is determined whether or not the sizing device 14 has a manual correction value transmitting function. Is determined. Since it has, the determination is YES, the presence or absence of manual correction in the sizing device 14 is determined in S67, and if not, the determination is NO.
Then, in S68, the final correction value U ^* is transmitted to the sizing device 14 as the automatic correction value U of this time, and is stored in the auxiliary storage device 22. Then, in S69, it is determined whether or not there is an auxiliary correction command. Since it is assumed that this time does not exist, the determination is NO, and in S70, the matching fuzzy rule is stored in the auxiliary storage device 22 as in S65.

Details of S65 and S70 will be described below with reference to the flowchart of FIG.

First, in S401, the current value of the abnormality counter is read from the RAM. Next, S4
In 02, the fuzzy rule (represented by "adaptation rule" in the figure) that is adapted in the fuzzy operation by the current execution of S55 of FIG. 7 is the abnormal fuzzy rule (represented by "abnormality rule" in the figure). It is determined whether or not. This time, if it is assumed that the abnormal fuzzy rule does not match, the determination becomes NO, and immediately the process proceeds to S404. However, this time, assuming that the abnormal fuzzy rule has met, the determination is YES,
After the value of the abnormality number counter is incremented by 1 in S403, the process proceeds to S404. This S4
In 04, the current value of the judgment number counter is read from the RAM, and thereafter, in S405, the value of the judgment number counter is incremented by one. Then, in S406, it is determined whether or not the current value of the determination number counter is larger than the set value B. If it is not large this time, the determination is NO, the latest values of the determination number counter and the abnormality number counter are stored in the RAM in S407, and then, in S408, the adapted fuzzy rules are stored in the auxiliary storage device. 22 is stored.

Thereafter, as a result of the execution of the routine shown in the figure being repeated many times, the current value of the judgment number counter is the set value B.
If it becomes larger, the determination in S406 becomes YES, and S406
In 409, the current value of the abnormal number counter is the set value C.
It is determined whether or not it is greater than. If it is not large this time, the determination is NO, and in S410,
The abnormality flag is turned off (that is, the abnormality flag is turned on.
Is changed from OFF to OFF or is maintained OFF), and thereafter, in S411, the values of the determination number counter and the abnormality number counter are reset to zero. After that, the process proceeds to steps S407 and below.
On the other hand, assuming that the current value of the abnormality counter is larger than the set value C, the determination in S409 is YES, and the abnormality flag is turned on in S412 (that is, the abnormality flag is changed from OFF to ON). Alternatively, it is maintained ON), and thereafter, the process proceeds to the steps after S411.

The case where there is no manual correction in the sizing device 14 has been described above. However, assuming that there is a manual correction, the determination in S67 of FIG. 9 becomes YES and S71.
In, the latest manual correction value and the sizing point are received and stored from the sizing device 14, the work wait flag is turned on in S72, the operation data memory is cleared in S73, and then the process returns to S8.

Assuming that the sizing device 14 does not have the manual correction value transmitting function, the determination in S66 is NO, and in S74, the latest sizing point is read from the sizing device 14, It is stored in the RAM and is also stored in the auxiliary storage device 22, and the previous sizing point is read from the RAM in S75. After that, in S76, it is determined from the previous sizing point and the latest sizing point whether or not the sizing point has been changed, that is, it is manually operated by a sizing device that does not have a manual correction value transmission function. It is determined whether or not there is a correction, and if there is no change, the determination is NO, and the process proceeds to S68, but if there is, the determination is YES, and in S77, the work waiting flag is turned ON and S
At 78, the operation data memory is cleared, and then the process returns to S8.

On the other hand, assuming that there is an auxiliary correction command, the determination in S63 is YES, and in S79, it is determined whether or not the auxiliary correction is being executed. It is determined whether or not the value of the auxiliary correction counter, which indicates the number of times of execution of the auxiliary correction, is 1 or more. If it is assumed to be 0 this time, the determination becomes NO, the process proceeds to the step group of S64 and below, and the main correction is performed. In S69 of this step group, it is determined whether or not there is an auxiliary correction command, and it is assumed that this is the case, so the determination is Y.
ES is reached, and the value of the auxiliary correction counter is incremented by 1 in S80.

On the other hand, assuming that the auxiliary correction is currently being executed and the value of the auxiliary correction counter is not 0, the determination in S79 is YES, and the process proceeds to the step group of S81 and thereafter. Is done. In S81, a value obtained by subtracting the last final correction value U ^* from the last final correction value U ^* is set as the current transmission value. Here, "this time of the final correction value U ^*", the equivalent to this time of the provisional correction value U _P, "the last of the final correction value U ^*" corresponds to a provisional correction value U _P of the last time, " The "current transmission value" corresponds to the current final correction value U _F. Then, in S82, it is determined whether or not the sizing device 14 has a manual correction value transmission function. Since the sizing device 14 has the manual correction value transmitting function, the determination is YES, and in S83, it is determined whether or not the sizing device 14 is subjected to the manual correction.
At, the transmission value is transmitted to the sizing device 14 as the automatic correction value U of this time. Auxiliary correction is performed.
Thereafter, in S85, the transmission value is stored in the auxiliary storage device 2
2, the auxiliary correction counter is incremented in S86, and then the process proceeds to S70. On the other hand, assuming that there is manual correction, the determination in S83 is Y.
The process becomes ES, the manual correction value is received from the sizing device 14 in S87, the work waiting flag is turned ON in S88, the operation data memory is cleared in S89, and then the process returns to S8.

On the other hand, assuming that the sizing device 14 does not have the manual correction value transmission function, the determination in S82 is NO, and the process proceeds to the step group of S74 and subsequent steps for automatic correction. Whether or not the permission is permitted is determined, and if the permission is permitted, the automatic correction value is transmitted to the sizing device 14. Also,
In this case, the determination in S69 is YES, and in S80, the auxiliary correction counter is incremented.

When the execution of S70 is completed, S9 of FIG.
At 0, the presence or absence of the auxiliary correction command is determined. Assuming that this time does not exist, the determination is NO, the work wait flag is turned on in S91, the calculation data memory is cleared in S92, and then, in S93, the presence or absence of the data shift processing command is determined. Since it is assumed that this time is not the case, the determination is NO, and the process immediately returns to S8.

On the other hand, assuming that there is an auxiliary correction command this time, the determination in S90 is YES, and in S94, it is determined whether or not this auxiliary correction should be ended. Specifically, it is determined whether or not the current value of the auxiliary correction counter is equal to or greater than the set value (the value input from the auxiliary storage device 22 in S1 of FIG. 5). If this is not the case this time, the determination is no and the process immediately returns to S8.

After that, assuming that the current value of the auxiliary correction counter becomes equal to or larger than the set value while the execution of this routine is repeated many times, the determination in S94 becomes YES,
In step S95, the sum of all automatic correction values U (hereinafter, referred to as "total correction value") transmitted to the sizing device 14 in at least the current auxiliary correction of the current auxiliary correction and the main correction preceding this is calculated. It Then S
At 96, it is determined whether or not the total correction value is 0, that is, whether or not the current auxiliary correction needs to be continued at least because it is estimated that the current auxiliary correction has not been performed at a time when it is really necessary. To be judged. If this is not the case this time, the determination is NO, the work wait flag is turned on in S97, the calculation data memory is cleared in S98, and the presence or absence of the data shift processing command is determined in S99. Since it is assumed that this time is not the case, the determination is NO, and the process immediately returns to S8.

On the other hand, assuming that it is necessary to continue the current auxiliary correction, the determination in S96 becomes YES, and in S100, it is determined whether or not the auxiliary correction restart command has been issued. If it is assumed that there is an auxiliary correction extension command instead of an auxiliary correction restart command this time, the determination is NO and S10
At 1, the value of the auxiliary correction counter is set to 1, and then the process returns to S8. Therefore, at the next execution of this routine, the current value of the auxiliary correction counter is not 0,
The determination in S79 of FIG. 9 is NO, and the process proceeds to S64.

On the other hand, if it is assumed that there is an auxiliary correction extension command this time instead of an auxiliary correction extension command, the determination in S100 of FIG. 10 is YES, and the value of the auxiliary correction counter is set to 0 in S102. Then, the process returns to S8.
Therefore, at the next execution of this routine, since the current value of the auxiliary correction counter is 0, the determination in S79 of FIG. 9 becomes YES, and the process proceeds to S81.

When S25 of FIG. 6 is executed with the work waiting flag turned on, the determination is YES.
In S103, the calculation data memory is cleared in S103, and then the process returns to S8. That is, immediately after the manual correction or the automatic correction of the sizing point, S in FIG.
Despite the existence of 23, the measured value X, etc. is not actually stored in the calculation data memory, and the work piece machined under the sizing point affected by the latest manual correction or automatic correction is the first. The work waiting flag is turned off when the measurement is performed by the total number measuring device 16, and the determination in S25 of FIG.
The value becomes O, and the accumulation of the measured value X and the like in the calculation data memory is restarted.

The case where the data shift processing command has not been issued has been described above. Next, the case where it has been issued will be described.

In this case, the determination in S24 of FIG. 6 is YE.
It becomes S, and the process proceeds to the step group after S104. S1
At 04, it is determined whether or not the data shift process should be prohibited. Regarding the top correction target work related to the latest manual correction value or automatic correction value
When the measurement by is finished, it is no longer necessary to perform the data shift processing, and if it is performed, the dimensional error of the work increases, so in such a case, the data shift processing is prohibited.

The determination as to whether or not the measurement of the top correction target work is completed is specifically performed by the target number of data shift processes stored in the RAM (the storage of this will be described later), that is, the latest number. The number of workpieces existing between the processing machine 10 and the total measuring machine 16 when the manual correction value or the automatic correction value is transmitted to the sizing device 14 is measured every time the total measuring machine 16 finishes the measurement. The value is subtracted by one, and when it becomes 0 as a result, it is determined that the measurement of the head correction target work is completed.

If it is assumed that the measurement of the work piece to be head-corrected has not been completed this time, the determination becomes NO and S10
In 5, the latest manual correction value or automatic correction value is determined as the current shift amount, and then in S106, the current measurement value X is changed by adding the current shift value to the current measurement value X. , And is stored in the auxiliary storage device 22 while being accumulated in the arithmetic data memory. Then, in S107, the past measured value X is input from the calculation data memory, and the preparation for the calculation of the moving average value P in S27 and thereafter is performed.

Thereafter, if it is determined in S93 of FIG. 10 whether or not there is a data shift processing command, it is assumed that this is the case, so the determination is YES, and the dead time MS from the work number counter 18 is determined in S108. It is input, stored in the RAM as the target number of the next data shift processing, and further stored in the auxiliary storage device 22. Then, the process returns to S8.

Further, similarly to the above case, the determination in S99 of the figure is also YES, and in S109, the dead time MS is input from the work number counter 18 and stored in the RAM as the target number of the next data shift process. Stored in the auxiliary storage device 22. Then, the process returns to S8.

As described above, the target number of data shift processes is performed with the end of the intermittent correction.
Although not shown in the figure, it is also performed when the manual correction is completed, and the standard value thereof is stored in the ROM in advance in preparation for the first execution of this routine.

After that, assuming that the measurement by the total number measuring machine 16 is completed for the head correction target work related to the latest manual correction value or automatic correction value while the execution of this routine is repeated many times, FIG. The determination in S104 is Y
ES is reached, the shift amount this time is set to 0 in S110, and then the process proceeds to S106. That is, this time, the actual measured value X is directly stored in the calculation data memory.

As described above, when the data shift process is permitted, the measured value X is stored in the calculation data memory regardless of the ON / OFF state of the work waiting flag, and as a result, the automatic correction interval is increased. The time gets shorter.

As is clear from the above description, in the present embodiment, the processing machine 10 constitutes one aspect of the "processing machine 1" in each of the inventions of claims 1 to 3, and the sizing device 14 and the motor controller are included. 15 constitutes one aspect of the "machining machine control means 2", the total number measuring instrument 16 constitutes one aspect of the "measuring instrument 3", and FIG.
The part that executes the steps other than S64, 68 and 84 of the above constitutes one mode of the "correction value determination means 4", and these S
The part that executes 64, 68 and 84 constitutes one mode of the “correction value transmission means 5”.

Although one embodiment of the present invention has been described in detail with reference to the drawings, the present invention can be implemented in other modes.

For example, in the above embodiment, the measured value X
As a method of measuring the vibration level of the above, a method of measuring based on the relationship between the measured value X and the moving average value P and a method of measuring based on the content of the fuzzy rule adapted in the fuzzy calculation have been adopted. All of these methods belong to the method of measuring the vibration level in the process of determining the correction value U, but other methods for measuring the vibration level in the process of determining the correction value U include the following. Things can be adopted.

For example, as a method of using the moving average value P, a method of measuring a standard deviation of a plurality of moving average values P as a vibration level, a method of measuring a vibration level as a power spectrum density at a setting frequency in question, and A method of measuring the vibration level by pattern recognition using a neural network, a correlation function, etc. can be adopted.
Further, those methods can be adopted when the correction value U is used instead of the moving average value P.

The vibration level can also be measured by using the measured value X itself and by a method according to some of the above-mentioned methods.

Further, the vibration level is measured by the sizing device 1
Correction value U sent to No. 4 and measurement value X affected by it
It can also be done based on the relationship with. The vibration level of the measured value X can also be measured based on the response of the measured value X to the correction value U.

Further, in the above embodiment, changing the rule for determining the correction value U makes the value of the weighting coefficient b relating to the moving average value P variable and the content of the fuzzy rule in the fuzzy calculation variable. However, it can be realized by either one of them or by another method. For example, the value of the number (= K) of the measurement values X necessary for calculating the moving average value P by the principle method is variable, or the order of the regression line assumed when calculating the moving average value P ( For example, the primary, secondary, etc.) can be made variable, or the differential value T
The number of moving averages P (= L) required to calculate is calculated by the principle method, the number and contents of labels of fuzzy variables are changed, and the characteristics of the membership function of fuzzy variables (for example, The peak value of the grade, the shape of the function, etc.) can be made variable, and the method (for example, the center of gravity method, the area method, etc.) when defuzzifying the fuzzy inference value to the actual value is made variable, and the final correction value U ^* It can also be realized by changing the number (= M) of provisional correction values U necessary for determination in consideration of the continuity thereof, or by changing the dead zone width N of the correction value U.

In the above embodiment, the weighting factor b
When the value and the fuzzy rule are changed, the correction value U is decided by using the data stored in the operation data memory before clearing the data as it is. This is because if the calculation data memory is cleared every time there is a change, a situation in which the accumulated data is insufficient will occur rather frequently, and it is expected that the dimensional accuracy of the workpiece will not be sufficiently stable. On the other hand, on the other hand, even if the correction value U is determined by using the data before the change without being excluded, the correction value U that is not so appropriate is not acquired. However, it is possible to implement the present invention in such a manner that the calculation data memory is cleared every time there is a change.

Further, in the above embodiment, the moving average value P, the error value R, the differential value T, etc. are successively calculated even when the number of the measured values X stored in the arithmetic data memory does not reach the set number. One final correction value U ^* is determined when the number of measured values X that have been calculated and stored in the calculation data memory has reached a set number.
However, when the number of accumulated measurement values X does not reach the set number, the moving average value P or the like is not calculated at all, and when the number of accumulated measurement values X reaches the set number. For the first time, the present invention can be implemented by collectively calculating the moving average values P and the like to determine one final correction value U ^* .

Further, in the above-mentioned embodiment, the crankshaft is used as the work, and a plurality of journal surfaces (outer peripheral cylindrical surface) of the work are used.
This is an example of the case where the present invention is applied to a sizing point correction device used together with a processing system that performs cylindrical grinding as each processing portion. However, the present invention is applied to a sizing point correction device used with other processing systems. Of course, it can be applied. Other machining systems include, for example, a cylinder block of an automobile engine as a workpiece to be machined,
It is possible to select a processing system in which a plurality of cylinder bores (inner circumferential cylindrical surface) formed in advance as the processing parts are honed.

Further, the above embodiment is also an example in which the present invention is applied to a machining system for machining a workpiece in which a plurality of machining parts are set, but machining in which only one machining part is set. Of course, it can be applied to the system.

Further, the above-mentioned embodiment is a machining system for machining a workpiece in which a plurality of machining parts are set, and the present invention is applied to those in which the in-process measuring machine is not provided for all the machining parts. This is also an example of the case, but it goes without saying that the present invention can be applied to a processing system including an in-process measuring machine for all of those processing parts.

Other than these, the present invention can be implemented in various modified and improved modes based on the knowledge of those skilled in the art without departing from the scope of the claims.

[Brief description of drawings]

FIG. 1 is a diagram conceptually showing the structure of each invention of claims 1 to 3.

FIG. 2 is a perspective view showing a state in which a crankshaft is ground by a grindstone in a processing system used with a feedback type sizing point correcting apparatus which is an embodiment common to the inventions of claims 1 to 3;

FIG. 3 is a system diagram showing the entire processing system.

FIG. 4 is a diagram showing a configuration of a processing machine in the processing system.

5 is a flowchart showing a part of a fixed-point correction routine executed by a computer of control device 20 in FIG.

FIG. 6 is a flowchart showing another part of the sizing point correction routine.

FIG. 7 is a flowchart showing still another part of the sizing point correction routine.

FIG. 8 is a flow chart showing still another part of the sizing point correction routine.

FIG. 9 is a flowchart showing still another part of the sizing point correction routine.

FIG. 10 is a flowchart showing still another part of the sizing point correction routine.

FIG. 11 is a diagram conceptually showing the overall flow of the processing of the sizing point correction routine.

12 is a graph showing the value of each weighting coefficient b used for removing the variation between adjacent portions in FIG. 11.

FIG. 13 is a graph conceptually showing how the value of each weighting coefficient b is automatically changed in the above embodiment.

FIG. 14 is a diagram for explaining the number of measurement values X required to calculate one moving average value P by the principle method in the above-mentioned embodiment.

FIG. 15 is a diagram for explaining the number of measurement values X required to calculate one moving average value P by the first exceptional method in the above-mentioned embodiment.

FIG. 16 is a diagram for explaining the number of measurement values X necessary for calculating one moving average value P by the second exceptional method in the above-mentioned embodiment.

17 is a graph conceptually showing the contents of both-ends diameter correction in FIG.

18 is a graph conceptually showing a method of calculating a differential value T from an error value R in obtaining the dimension information in FIG.

19 is a graph showing a membership function used for the error value R in the fuzzy calculation in FIG.

FIG. 20 is a graph showing a membership function used for a differential value T in the fuzzy operation.

FIG. 21 is a graph showing a membership function used for the correction value U in the fuzzy calculation.

22 is a graph conceptually showing the content of consideration of continuity in FIG.

FIG. 23 is a diagram for explaining an example of a process of deriving a final correction value U ^* from the measurement value X in the sizing point correction routine of FIGS.

FIG. 24 is a graph conceptually showing the content of dead zone consideration in FIG. 11.

FIG. 25 is a graph conceptually showing intermittent correction which is one method of fixed-point correction.

FIG. 26 is a graph conceptually showing the first intermittent correction method.

FIG. 27 is a graph conceptually showing one embodiment of the first system.

FIG. 28 is a graph conceptually showing the second intermittent correction method in FIG. 25.

FIG. 29 is a graph conceptually showing one embodiment of the second method.

FIG. 30 is a flowchart showing details of S35 in FIG.

FIG. 31 is a flowchart showing details of S55 in FIG.

32 is a flowchart showing details of S65 and S70 in FIG. 9. FIG.

[Explanation of symbols]

 10 processing machine 12 in-process measuring machine 14 sizing device 15 motor controller 16 total number measuring machine 20 controller 44 post-process measuring machine 50 keyboard

Claims (3)

[Claims] 1. A processing machine for sequentially processing each of a plurality of workpieces, and (b) a processing condition of the processing machine is determined based on a correction value supplied from the outside, and according to the determined processing condition. Processing machine control means for controlling the processing machine, and (c) those processing machine control means and measuring machine of a processing system including a measuring machine for sequentially measuring the dimensions of each of a plurality of workpieces processed by the processing machine A feedback type machining condition correction device to be used by being connected to the machining machine according to a rule that changes based on a measurement value by the measurement machine and a vibration level which is a vibrational variation level of the measurement value. A correction value determining means for determining a correction value of the processing condition of the work to be processed next, and a correction value supplying means for supplying the determined correction value to the processing machine control means. Fed back type processing conditions correction device. 2. The correction value determining means measures the vibration level of the measured value, selects one that matches the measured vibration level from a plurality of preset rules, and follows the selected rule. The feedback type processing condition correction device according to claim 1, wherein the correction value is determined. 3. The correction value determining means measures a vibration level of the measured value, corrects a preset rule based on the measured vibration level, and determines the correction value according to the corrected rule. The feedback type processing condition correcting device according to claim 1, which is a device.