JPH04159077A - Grinding machine control device - Google Patents

Abstract

PURPOSE:To perform computation of a variation point in a short time by automatically controlling the variation point of a constant cutting speed through fuzzy inference, based on a detecting value for a polishing state by a sensor. CONSTITUTION:A sensor 7 to detect the grinding state of a material 6 to be ground and a control means 11 to control the variation point of a constant cutting cut speed, based on an output signal from the sensor 7. Based on a detecting value for a polishing state by the sensor 7, a plurality of the variation points of a constant cut speed are automatically controlled through fuzzy inference by means of a fuzzy inference device 10. Namely, when cutting is effected at a specified speed, the variation point is automatically determined from a detecting value for a polishing state through fuzzy inference. During arrival at the variation point, cutting is effected at a subsequent constant speed, and during the occurrence of spark out, a cut speed is reduced to zero.

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention relates to a grinding machine control device, and in particular, - automatically changes the constant cutting speed for each workpiece by applying fuzzy reasoning. It relates to a device to be controlled.

<Conventional technology> Conventionally, in the grinding cycle of a cylindrical grinder, the constant cutting speed of the grinding wheel was changed from rough grinding (coarse grinding) to precision grinding (seiken), and from fine grinding to spark-out. In both cases, the values were fixed among the same type of workpiece.

Since the point at which the cutting speed of the grindstone is changed is a fixed value, the following disadvantages (1) to (4) have occurred.

■ There is a limit to reducing machining time. This is especially noticeable when the change from rough grinding to fine grinding is fixed.

■ There are variations in quality.

■ Variations in processing accuracy occur.

These reasons are as follows (jl~(■)).

The sharpness of the fit whetstone is good immediately after dressing, but it gradually deteriorates due to crushing and clogging. (ii) With the cutting speed change point fixed method, if the cutting speed is good, the change point from coarse to fine grinding or from fine grinding to spark out will be reached quickly, and the finishing dimension ( (finished diameter and finished width)
Although it takes a short time to reach this point, the cut will be excessive, and the finished dimension will not be reached smoothly, resulting in poor quality (surface roughness).

- On the other hand, if the cutting is poor, the surface roughness will be improved, but there will be uncut parts and the time required to reach the finished dimensions will be longer.

− In this way, depending on the sharpness, excessive depth of cut or uncut parts may occur, resulting in variations in machining accuracy and variations in surface roughness. , the points at which the constant cutting speed changes, such as the spark-out starting point, are set to larger fixed values.

M As a result, the machining time becomes longer and varies greatly depending on the degree of sharpness. Moreover, if the sharpness is so poor that a large uncut portion is left, an auxiliary pick is required, and the time required to reach the finished diameter or finished width becomes even longer.

On the other hand, adaptive pure grinding using a cylindrical grinder is known as disclosed in Japanese Patent Publication No. 60-3951.

With this technology, the grinding situation is modeled in advance using mathematical formulas, and by inputting the actual measured value of the workpiece diameter, etc. into the control device, the change point of the constant cutting speed can be calculated during the grinding process. was. However, in this adaptive monument grinding,
Inconveniences such as the following fat~(b) have occurred.

ta+ Grinding conditions change from moment to moment depending on the condition of the workpiece, the condition of the grindstone, etc. Therefore, if a grinding situation is to be modeled strictly, an extremely large number of parameters will be required, and the calculation time will increase. Due to this increase in calculation time, there is a limit to the length of the sanburi leg interval and it becomes longer, and as a result, it is not possible to respond to changes in the fine grinding situation.

(b) On the other hand, if the grinding situation is simplified and modeled, the accuracy of adaptation to each work material will deteriorate.

<Problems to be Solved by the Invention> The present invention builds on the above-mentioned conventional technology, and automatically determines the change point of the constant cutting speed of the grinding wheel for each workpiece, thereby improving machining accuracy and stabilizing quality. The object of the present invention is to provide a grinding machine control device that can achieve all of the following:

Means for Solving the Problems> The configuration of the present invention includes a grindstone that is rotationally driven by a drive device, a headstock that holds a workpiece, and at least one of the grindstone and the headstock that is moved at a plurality of constant speeds. A grinding machine control device comprising: a moving mechanism that makes a cut; a sensor that detects the grinding state of a workpiece; and a control means that controls a point at which a constant cutting speed is changed based on an output signal of the sensor. Based on the value detected by the sensor on the grinding state, multiple constant cutting speed changes are automatically determined using fuzzy inference. The invention is characterized by comprising a fuzzy inference device that performs Im. However, the constant speed can also include speed 4.

When cutting is performed at a constant speed, the points of change are automatically determined by fuzzy reasoning from the detected values of the grinding state. When this change point is reached, the cut is made at the next constant speed. At the time of spark-out, the cutting speed is assumed to be zero.

Embodiments Hereinafter, the present invention will be described along with embodiments with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of the apparatus according to the first embodiment, in which a workpiece size detector 7 and a motor load detector 9 are used to calculate the motor load and uncut amount as detected values of the grinding state. By inputting the data to the fuzzy inference device 10, fuzzy inference as shown in FIGS. 2 to 9 is performed to control changes in the constant cutting speed.

In FIG. 1, in a cylindrical grinding machine main body 1, a workpiece 6 is held by a headstock 2 or a headstock 2 and a tailstock 3, and is also rotationally driven.

Further, the grindstone 5 is rotationally driven by the drive motor 4. Further, the headstock 2 is moved together with the tailstock 3 by a moving mechanism (not shown) to adjust the depth of cut. Of course, the amount of cut may be adjusted by moving only the grindstone 5 or both the grindstone 5 and the headstock 2 using a moving mechanism.

The cutting speed by the moving mechanism is changed by a command from the control device 12, and remains constant until the changing point is reached, and then becomes the next constant speed.

A workpiece size detector 7 and a motor load detector 9 are attached to the grinding machine body 1.

The workpiece size detector 7 is a contact or non-contact device that actually measures the dimensions of the workpiece 6 during machining (for example, the diameter for plunge, traverse, and taper grinding, and the radius for shoulder grinding). Automatic mold sizing is used.

Detection signal 7A output from this work material size detection I#7
and the momentary grinding wheel set cutting amount 11A calculated from the NC command value by the NC device 11 are input to the subtraction circuit 8 in the control value [12, and this subtraction circuit 8 calculates the uncut amount 8.
A is calculated and input to the fuzzy inference device 10.

The motor load detection N9 detects the motor load during grinding by detecting the load current of the motor 4 during grinding, and inputs the detection signal 9A to the fuzzy inference device 10. Note that, in addition to the grindstone drive motor 4, the motor to be subjected to load detection may be a motor that rotationally drives the workpiece 6.

The fuzzy inference device 10 calculates the optimal cutting speed change point for the grinding condition based on the input values (motor load, uncut amount) representing the grinding condition, and the result (data) 1
0A is given to the NC device 1-11.

The hypothetical calculation is always performed during cutting at a constant speed (temporarily assumed to be the first constant cutting speed), and the NC device 11 calculates the cutting speed change point, which is the fuzzy inference result, and
Input the actual dimension value measured by the work material dimension detector 7, and when these values match, give the speed switching command 11B to the grinding machine body 1 to start the next constant speed (second constant cutting speed). change it to

As a result, the grinding machine main body 1 performs cutting at a predetermined second constant cutting speed, and during this cutting, the fuzzy inference device 10 infers the next cutting speed change point.

In addition, the subtraction circuit 8, the fuzzy inference device [10] and the NC device [1]
1 may have the same device configuration.

For safety, if the output of the motor load detection I19 exceeds the reference value (usually 120%), the output of the motor load detection I#9 is directly output to the NC without passing through the fuzzy inference device 10.
It is also possible to input the information to the vicl 111 and take appropriate safety measures.

Furthermore, as the set cutting depth, an actual value measured using cutting speed detection may be used instead of the NC command value.

Next, the fuzzy inference device 10 will be explained in detail. Here, as an example, fuzzy inference is performed using two human power and one output, and the optimal cutting speed change point from coarse grinding to fine grinding is output based on the membership functions shown in Figures 2 to 4. Furthermore, the present invention is configured to output the optimum cutting speed change point from fine cutting to sparking out based on the membership functions shown in FIGS. 6 to 8.

First, the fuzzy inference of the cutting speed change point from coarse grinding to fine grinding will be explained. FIG. 2 shows an example of a membership function in which the input value is the grindstone motor load X from the motor load detection M9. FIG. 3 shows an example of a membership function whose input value is the uncut amount x2 obtained from the measured dimension value from the work material dimension detector 7 and the set cutting depth calculated by the NC device 11. FIG. 4 shows an example of a membership function whose output value is the cutting speed change point Y1 (expressed as the difference from the finished dimension) from rough grinding to fine grinding. In addition, some of the control rules used in this example are the following (a) and (b), and the fifth
The figure shows the matrix of each barb rule.

k) If the grindstone motor load x1 is large (β) and the uncut amount X2 is small (S+), the cutting speed change point Y1 from rough grinding to fine grinding is made very large (VB).

(b) If the grindstone motor load x1 is small (Sl and the uncut amount x2 is large), the cutting speed change point Y1 from rough grinding to fine grinding is made very small (VS).

Next, fuzzy inference regarding the cutting speed change point from fine grinding to spark out will be explained. Basically, Figure 2~
This is the same as the case in FIG. 5, and FIG. 6 shows an example of a membership function using the grindstone motor load x1 from the motor load detector 9 as an input value. FIG. 7 shows an example of a membership function whose input value is the uncut amount x2 obtained from the actual dimension value from the work material dimension detector 7 and the set cutting amount calculated by the NC device f1. Figure 8 shows cutting speed change point Y from fine grinding to spark out (expressed as difference from finished dimension)
Here is an example of a membership function whose output value is . Also,
Some of the control rules used in this example are as follows (a), (
(b), and FIG. 9 shows the matrix of each splinter rule.

(B) If the grindstone motor load x1 is large (C) and the uncut amount X2 is small (S), the cutting speed change point Y1 from fine grinding to spark out is made very large (VB).

(b) If the grinding wheel motor load x1 is small and +31 and the uncut amount x2 is large (β), then the cutting speed change point Y1 from fine grinding to spark out should be made very small (
VS).

The Mediburn Tube functions as shown in FIGS. 2 to 4 and 6 to 8 described above are defined as for each side above for each constant cutting speed that has already been determined. Alternatively, they can be defined on the same table.

Further, as the input value signal to the fuzzy inference device 10, in addition to the two manual force values of the grindstone motor load and the amount of uncut material shown in FIG. 1, any signal that represents the grinding state may be used. An example is shown in FIG.

In the example shown in FIG. 10, the grinding machine body 1 includes a workpiece size detector 7, a motor load detector 9, and a cutting speed detector @1.
4. Contact spark detector 15 between grinding wheel 5 and workpiece 6, surface texture (chatter mark) detector 16, grinding sound detector 17, A
E wave detector 18, surface roughness detector 19, vibration detection M20
, a fi degree detector 21 are attached, and the detected values of the grinding state by these are inputted to the fuzzy inference device 10. However, any number of detected values may be used, and fuzzy inference may be performed using any combination of detected values. - As an example, refer to Figs. 10 to 14 for an example of using fuzzy inference to determine the optimal change in cutting speed using the amount of change in workpiece dimensions (e.g., diameter) and the amount of uncut material as input values. I will explain.

In FIG. 10, the differential circuit 22 is connected to the workpiece size detector 7.
The difference between the measured diameter value from and the measured diameter value past a certain period of time, that is, the amount of change in diameter of the workpiece (temporal change) is determined and the signal 22A is input to the fuzzy inference device 10. The subtraction circuit 8 calculates the uncut amount from the actual diameter value from the workpiece size detection #7 and the cutting amount of the grindstone based on the cutting speed detector 14, and inputs the signal 8A to the fuzzy inference system [10].

FIG. 11 shows an example of a membership function whose input value is the amount of change in workpiece diameter x1 from the difference circuit 22. FIG. 12 shows an example of a membership function whose input value is the uncut amount X- obtained from the subtraction cycle @8. Figure 13 shows cutting speed change point Y from fine grinding to spark out.

An example of a membership function whose output value is (expressed as the difference from the finished dimension) is shown below. Further, some of the control rules used in this embodiment are as shown in (a) and (b) below, and FIG. 14 shows a matrix of each control rule.

(B) If the amount of workpiece diameter change X1 is large and CB)
Moreover, if the uncut amount x2 is small (31), the cutting speed change point Y1 from fine grinding to spark out is made very large (VB).

(b) If the work material diameter change amount X1 is small (s)
Moreover, if the uncut amount X2 is large (v), the cutting speed change point Y from fine sharpening to spark out is made very small (VS).

Note that the number of change points may be one or more, and any number of changes may be calculated.

When using something other than the grindstone motor load, the amount of uncut material, and the amount of dimensional change of the workpiece as the input value signal, the following will occur.

Cutting speed: The cutting speed commanded by the NC system [1] and the actual cutting speed of the grinding machine body 11 do not necessarily match, and differ depending on the grinding state.

Therefore, by actually measuring the cutting speed with the cutting speed detector 14 and using this as an input value for fuzzy inference, accurate control can be performed.

Surface texture: When light is applied to the workpiece 6, the reflected light may draw a striped pattern (chatter marks). If such chatter marks occur, the value of the product may decrease even if the dimensional accuracy and surface roughness meet the required values. Therefore, by detecting the surface texture using the surface texture detection M16 and using this as an input value for fuzzy inference, it is possible to control grinding with good quality.

Surface roughness: Since the product has a required surface roughness, the current surface roughness of the workpiece 6 is detected by the surface roughness detector 19 during grinding, and when the surface roughness reaches the required value, The whetstone 5 is relatively separated from the workpiece 6 and the Tesshake-out is stopped. This makes it possible to obtain the required accuracy in a short period of time without causing unnecessary spark-outs. Furthermore, the detected value of surface roughness is used as an input value for fuzzy inference, if necessary.

Vibration: The generation of vibration during machining is related to the rigidity of the grinding wheel 5, the rigidity of the workpiece 6, etc., but vibration also causes chatter marks and the like. Therefore, the grinding state can be grasped by detecting the vibrations using the vibration detection pin 20, and control with better quality than the input value of fuzzy inference can be achieved.

Temperature: The temperature of the workpiece 6 during processing is related to the quality of the workpiece 6, such as its surface hardness. Furthermore, the temperature of the grinding oil, lubricating oil, etc. in the grinding machine body 1 also affects the grinding state. Therefore, temperature detection M2
By using the detected value of 1 as the input value of fuzzy inference,
Have good quality control.

AE wave: When the AE wave during processing is detected by the AE wave detector 18,
Even when it is difficult to detect contact between the grinding wheel 5 and the workpiece 6, such as during spark-out, the grinding state can be grasped. Therefore, A.E.
By using the detected wave values as input values for fuzzy inference,
Have good quality control.

Grinding sound detector: Skilled worker: Detects the machinability of the workpiece 6 and the grinding wheel 6 by the grinding sound and the appearance of sparks due to contact between the grinding wheel 5 and the workpiece 6.
Judgments are made by comprehensively evaluating the grindability, etc. Therefore,
Grinding conditions can be grasped by detecting grinding sounds and sparks using the grinding sound detector 17 and contact spark detection N15 and investigating their properties, and high-quality control can be achieved by using these detected values as input values for fuzzy inference. .

<Effects of the Invention> According to the present invention, changes in the constant cutting speed are automatically controlled by fuzzy reasoning based on the detected value of the grinding state. Therefore, unlike conventional adaptive splinter grinding, there is no need to strictly model the grinding situation, and changes can be calculated in a short time. Furthermore, it can respond to small changes in the grinding situation, which further improves machining accuracy, stabilizes quality, and shortens machining time.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention; FIGS. 2, 3, 4, 6, 7, 8, 1
1, 12, and 13 are diagrams showing membership functions, respectively; FIGS. 5, 9, and 14 are diagrams showing control rule matrices, respectively; and FIG. 10 is a diagram showing another implementation of the present invention. FIG. 2 is a block diagram showing an example configuration. In the drawing, 1 is the grinding machine body, 2 is the headstock, 3 is the Shinshindai, and 4
is a grindstone drive motor, 5 is a grindstone, 6 is a workpiece, 7 is a workpiece size detector, 8 is a subtraction circuit, 9 is a motor load detection board, 1
0 is a fuzzy reasoning device, 11 is an NC device, 12 is a control device, 14 is a cutting speed detector, 15 is a contact spark detector, 16 is a surface texture (chatter mark) detector, 17 is a grinding sound detector, 18 19 is an AE wave detector, 19 is a surface roughness detector, 20 is a vibration detector, 21 is a temperature detector, and 22 is a differential circuit. Patent applicant Mitsubishi Heavy Industries, Ltd. Agent

Claims (1)

[Scope of Claims] A grindstone that is rotationally driven by a drive device, a headstock that holds a workpiece, a moving mechanism that moves at least one of the grindstone and the headstock to make cuts at a plurality of constant speeds, and a workpiece. A grinding machine control device comprising a sensor that detects a grinding state of a material to be cut, and a control means that controls a change point of a constant cutting speed based on an output signal of the sensor, A grinding machine control device comprising: a fuzzy inference device that automatically controls a plurality of constant cutting speed changes using fuzzy inference.