JP2023122566A - Numerical control device, numerical control method, and storage medium - Google Patents

Abstract

To provide a numerical control device that can accurately suppress errors in the movement trajectory of a control object based on a command even when an acceleration/deceleration filter is applied to the command trajectory of the control object, a numerical control method, and a storage medium storing a numerical control program.SOLUTION: A numerical control device applies a filter to a command trajectory that is a movement trajectory of a control object moving according to a command. The numerical control device determines a corresponding point on a filter trajectory to which the filter is applied as a point corresponding to each of base points that are points on the command trajectory (S31). The numerical control device calculates, for each base point, an error vector indicating the distance and direction from the base point to the corresponding point (S33). The numerical control device calculates a gain corresponding to the calculated error vector for each base point (S37). The numerical control device applies the calculated error vector and gain to the corresponding base point and corrects the command trajectory (S41).SELECTED DRAWING: Figure 19

Description

The present invention relates to a numerical controller, a numerical control method, and a storage medium.

A numerical controller applies an acceleration/deceleration filter to a command signal for movement control of a tool or table (hereinafter referred to as "controlled object"). The numerical controller controls the movement of the controlled object by the command signal after applying the acceleration/deceleration filter, thereby suppressing the motor for moving the controlled object from being driven under excessive driving conditions. The movement trajectory of the controlled object that moves based on the command signal after application of the acceleration/deceleration filter may deviate from the movement trajectory of the controlled object that moves based on the command signal before application of the acceleration/deceleration filter. Techniques for suppressing this deviation have been proposed.

In the numerical control device described in Patent Document 1, first, it corresponds to the approximate value of the error between the reference machine axis path obtained from the transfer pulse before the acceleration/deceleration process and the machine axis path obtained from the transfer pulse after the acceleration/deceleration process. Calculate the amount of correction to be made. Next, the numerical controller moves the reference mechanical axis path by the correction amount in the direction of canceling out the error, and calculates the converted mechanical axis path. Next, the numerical controller converts the post-conversion machine axis path into an NC program to generate the post-conversion NC program. The numerical controller suppresses deviation between the reference machine axis path and the actual machine axis path by performing acceleration/deceleration processing based on the converted NC program.

Japanese Patent No. 5399537

The trajectory of the controlled object before application of the acceleration/deceleration filter is corrected using the error itself between the trajectory of the controlled object before application of the acceleration/deceleration filter and the trajectory of the controlled object after application of the acceleration/deceleration filter. In this case, there is a problem that residual error is large. The size of the remaining error at this time varies depending on the shape of the movement locus and the movement speed. In addition, in order to obtain the error between the movement trajectory of the controlled object before application of the acceleration/deceleration filter and the movement trajectory of the controlled object after application of the acceleration/deceleration filter, command points on the movement trajectory before and after application of the acceleration/deceleration filter must be associated. However, there is a problem that the calculation becomes difficult depending on the movement trajectory and the configuration of the acceleration/deceleration filter when the correlation is performed by the delay time of the acceleration/deceleration filter.

An object of the present invention is to provide a numerical control device, a numerical control method, and a numerical control program that can accurately suppress errors from the movement trajectory of a controlled object based on commands even when an acceleration/deceleration filter is applied to the movement trajectory of the controlled object. It is to provide a storage medium that stores.

A numerical control apparatus according to a first aspect of the present invention includes an acquisition unit that acquires a command for moving a controlled object that is at least one of a table and a tool of a machine tool, and the control unit that uses the command acquired by the acquisition unit. a first filtering unit that applies a first filter to a commanded trajectory that is a movement trajectory when an object moves; a determination unit that determines corresponding points on the filter trajectory that is the result of applying the first filter by the first filtering unit; A vector calculation unit that calculates an error vector indicating a distance and a direction for each base point; a gain calculation unit that calculates a gain corresponding to the error vector calculated by the vector calculation unit for each base point; a trajectory correction unit that corrects the command trajectory by applying the error vector calculated by the unit and the gain calculated by the gain calculation unit to the corresponding base points. .

The numerical controller calculates an error vector and a gain for each base point on the command locus. The numerical controller applies the calculated error vector and gain to each base point to correct the commanded trajectory. In this case, the numerical controller can optimize the gain for each base point on the command locus. Therefore, the numerical controller can approximate the movement trajectory of the controlled object when the first filter is applied to the movement trajectory of the controlled object when the first filter is not applied. Therefore, when applying the first filter to the commanded trajectory to control the movement of the controlled object, the numerical controller can accurately suppress the error between the actual movement trajectory of the controlled object and the original commanded trajectory.

In the first aspect, the command specifies at least two positions of the object to be controlled, and between the at least two positions specified by the command acquired by the acquisition unit, the command velocity acquired by the acquisition unit is used. An interpolating unit that interpolates, as the base point, a point that the machine tool passes through every control cycle when the object to be controlled moves, wherein the first filtering unit passes through the base point interpolated by the interpolating unit. The first filter may be applied to the commanded trajectory to be set. Hereinafter, the cycle (=control cycle) in which the interpolator performs interpolation will be referred to as an interpolation cycle. A numerical controller interpolates a plurality of base points between at least two positions defined by the command. Therefore, since the numerical controller can correct the commanded trajectory using the control points output in each actual control cycle, the error between the actual movement trajectory of the controlled object and the original commanded trajectory can be accurately corrected. can be suppressed.

In the first aspect, a distance calculation unit is further provided for calculating a movement distance in a tangential direction at the reference point interpolated by the interpolation unit based on the commanded trajectory, and the first filtering unit calculates the distance calculated by the distance calculation unit. The first filter is applied to the one-dimensional trajectory based on the moving distance, and the determining unit uses the one-dimensional trajectory based on the moving distance to which the first filter is applied to determine the corresponding one-dimensional trajectory corresponding to the base point. Corresponding points may be determined. The correspondence relationship between the base point interpolated from the one-dimensional command trajectory by the interpolator and the points on the one-dimensional filter trajectory that is the result of applying the first filter is used to associate the base points on the command trajectory with points on the filter trajectory. By doing so, the numerical controller can accurately determine the corresponding point on the filter trajectory corresponding to the base point on the command trajectory.

In the first aspect, an angular velocity calculation unit that calculates the angular velocity at the reference point interpolated by the interpolation unit based on the command locus, and the gain calculation unit calculates the angular velocity based on the angular velocity calculated by the angular velocity calculation unit. A gain may be calculated. The numerical controller can calculate the optimum gain for each base point on the command locus based on the angular velocity.

In the first aspect, the gain calculator calculates a function defined based on the relationship between the gain and the angular velocity that allows the filter trajectory to match the command trajectory when the command trajectory is circular. , the gain may be calculated for each angular velocity by applying it to the angular velocity calculated by the angular velocity calculator. The numerical controller can calculate the optimum gain for each base point on the command locus regardless of the value of the angular velocity.

In the first aspect, the function may represent an envelope of a graph in which the gain is plotted for each angular velocity. The numerical controller can accurately calculate the optimum gain for each base point on the command locus.

In the first aspect, further comprising a second filtering unit that applies a second filter having the same time constant as the first filter to the angular velocity calculated by the angular velocity calculating unit, wherein the gain calculating unit The gain may be calculated based on the angular velocity after the second filter is applied by the second filtering unit. The numerical controller can calculate the optimum gain even when the instantaneous change in angular velocity is large.
In the first aspect, a third filtering unit that applies the first filter to points on the corrected trajectory corrected by the trajectory correcting unit; a speed calculation unit that calculates the speed, a speed determination unit that determines whether or not the speed calculated by the speed calculation unit exceeds the command speed, and if the speed calculated by the speed calculation unit exceeds the command speed, the speed A speed changing unit may be further provided for changing the speed calculated by the speed calculating unit to be equal to or less than the command speed when the determining unit makes the determination. Since the numerical controller controls the controlled object at a speed equal to or lower than the commanded speed, it is possible to prevent deterioration of machining accuracy.

A numerical control method according to a second aspect of the present invention includes an acquiring step of acquiring a command for moving a controlled object, which is at least one of a table and a tool of a machine tool; A first filtering step of applying a first filter to a commanded trajectory that is a movement trajectory when an object moves; a determination step of determining corresponding points on the filter locus resulting from applying the first filter by the first filtering step; and an error indicating the distance and direction from each of the base points to each of the corresponding corresponding points. A vector calculation step of calculating a vector for each base point; a gain calculation step of calculating a gain corresponding to the error vector calculated by the vector calculation step for each base point; and the error calculated by the vector calculation step. a trajectory correction step of correcting the commanded trajectory by applying the vector and the gain calculated by the gain calculation step to the corresponding base point. In this case, the same effects as in the first mode are obtained.

A storage medium according to a third aspect of the present invention comprises: an acquisition step of acquiring a command for moving a controlled object, which is at least one of a table and a tool of a machine tool; A first filtering step of applying a first filter to the commanded trajectory, which is the movement trajectory when moving, and a point corresponding to each of the base points, which are points on the commanded trajectory, for the commanded trajectory a determination step of determining corresponding points on the filter trajectory resulting from applying the first filter by a first filtering step; and an error vector indicating the distance and direction from each of the base points to each of the corresponding corresponding points. a vector calculating step of calculating for each base point; a gain calculating step of calculating a gain corresponding to the error vector calculated by the vector calculating step for each base point; and the error vector calculated by the vector calculating step and a trajectory correction step of correcting the commanded trajectory by applying the gain calculated in the gain calculation step to the corresponding base point. In this case, the same effects as in the first mode are obtained.

1 is a perspective view of a machine tool 1; FIG. 2 is a block diagram showing electrical configurations of a numerical controller 30 and a machine tool 1; FIG. FIG. 10 is a diagram showing a command trajectory Gc(1) and a filter trajectory Gf(1); FIG. 10 is a diagram showing a command locus Gc(2); 5 is an enlarged view of the range W1 to W3 of FIG. 4; FIG. FIG. 4 is a diagram showing a multidimensional command trajectory Gc and a filter trajectory Gf; FIG. 4 is a diagram showing a one-dimensional command trajectory Dc and a filter trajectory Df; It is a graph which shows the relationship between the moving distance in multi-dimensions, and the variables i and j. It is a graph which shows the relationship between the moving distance and variables i and j in one dimension. 5 is an enlarged view of the range W1 to W3 of FIG. 4; FIG. FIG. 10 is a diagram showing a command locus Gc(3) and a filter locus Gf(3); 7 is a graph showing the relationship between the amount of change in the command locus Gc and the correction gain k; FIG. 10 is a diagram showing a command locus Gc(4) and a filter locus Gf(4); 7 is a graph showing an optimum solution for correction gain k calculated based on cumulative trajectory error; 7 is a graph showing a command trajectory Gc(5) and the angular velocity of the command trajectory Gc(5); 4 is a graph showing a correction gain function K(ω); 4 is a graph showing an example of a weighting function W(t) of an angular velocity filter; It is a flow chart of main processing. FIG. 18 is a flowchart of the main processing, continued from FIG. It is a photograph which shows the state on the base 2 of the machine tool 1 evaluated. Fig. 4 is a graph showing co-advance frequencies detected by an accelerometer; 4 is a graph showing the characteristics of an angular velocity filter; 4 is a table showing measurement results of cumulative trajectory errors; It is a figure which shows command locus|trajectory Gc(6) and filter locus|trajectory Gf(6), Gf(6)'. 7 is a graph showing respective errors of filter trajectories Gf(6) and Gf(6)'; The velocity of the controlled object moving along the command locus Gc(6) and the filter locus Gf(6), Gf(6)' is shown. 7 is a graph showing analysis results of vibration frequencies of beams; It is a figure which shows command locus|trajectory Gc(7) and filter locus|trajectory Gf(7), Gf(7)'. 7 is a graph showing respective errors of filter trajectories Gf(7) and Gf(7)'; The velocity of the controlled object moving along the command trajectory Gc(7) and the filter trajectories Gf(7) and Gf(7)' are shown. FIG. 10 is a diagram showing a command trajectory Gc(8) and filter trajectories Gf(8) and Gf(8)'; 7 is a graph showing respective errors of filter trajectories Gf(8) and Gf(8)'; The velocity of the controlled object moving along the command trajectory Gc(8) and the filter trajectories Gf(8) and Gf(8)' is shown. It is a flow chart of main processing of a modification. FIG. 34 is a flow chart of the main processing of the modification, continued from FIG. 34 .

Embodiments of the present invention will be described. In the following description, left and right, front and back, and top and bottom indicated by arrows in the drawings are used. The horizontal direction, the longitudinal direction, and the vertical direction of the machine tool 1 are the X-axis direction, the Y-axis direction, and the Z-axis direction of the machine tool 1, respectively. A machine tool 1 shown in FIG. 1 is a machine that rotates a tool 4 mounted on a spindle 9 and cuts a workpiece 3 held on the upper surface of a table 13 . A numerical controller 30 (see FIG. 2) controls the operation of the machine tool 1 .

<Overview of machine tool 1>
The structure of the machine tool 1 will be described with reference to FIG. The machine tool 1 includes a base 2, a column 5, a spindle head 7, a spindle 9, a table device 10, a tool changer 20, a control box 6, an operation panel 15 (see FIG. 2), and the like. The base 2 is a substantially rectangular parallelepiped base made of metal. A column 5 is fixed to the upper rear portion of the base 2 . A spindle head 7 is provided movably along the front surface of the column 5 in the Z-axis direction. The spindle head 7 rotatably supports the spindle 9 therein. The main shaft 9 has a mounting hole (not shown) in its lower part. The tool 4 is mounted in the mounting hole of the spindle 9, and is rotated by driving the spindle motor 52 (see FIG. 2). A spindle motor 52 is provided in the spindle head 7 . The spindle head 7 is moved in the Z-axis direction by a Z-axis movement mechanism (not shown) provided on the front surface of the column 5 . The numerical controller 30 controls the movement of the spindle head 7 in the Z-axis direction by controlling the drive of the Z-axis motor 51 (see FIG. 2).

The table device 10 includes a Y-axis movement mechanism (not shown), a Y-axis table 12, an X-axis movement mechanism (not shown), a table 13, and the like. The Y-axis movement mechanism is provided on the front side of the upper surface of the base 2, and includes a Y-axis rail, a Y-axis ball screw, a Y-axis motor 54 (see FIG. 2), and the like. The Y-axis rail and the Y-axis ball screw extend in the Y-axis direction. The Y-axis rail guides the Y-axis table 12 on the upper surface in the Y-axis direction. The Y-axis table 12 is formed in a substantially rectangular parallelepiped shape, and has a nut (not shown) on the bottom outer surface. The nut threads onto the Y-axis ball screw. When the Y-axis motor 54 rotates the Y-axis ball screw, the Y-axis table 12 moves along the Y-axis rail together with the nut. That is, the Y-axis moving mechanism supports the Y-axis table 12 so as to be movable in the Y-axis direction.

The X-axis movement mechanism is provided on the upper surface of the Y-axis table 12 and includes an X-axis rail (not shown), an X-axis ball screw (not shown), an X-axis motor 53 (see FIG. 2), and the like. The X-axis rail and the X-axis ball screw extend in the X-axis direction. The table 13 is formed in a rectangular plate shape in plan view, and is provided on the upper surface of the Y-axis table 12 . The table 13 has a nut (not shown) on its bottom. The nut screws onto the X-axis ball screw. When the X-axis motor 53 rotates the X-axis ball screw, the table 13 moves along the X-axis rail together with the nut. That is, the X-axis moving mechanism supports the table 13 so as to be movable in the X-axis direction. The table 13 is moved on the base 2 in the X-axis direction and the Y-axis direction by the Y-axis movement mechanism, the Y-axis table 12, and the X-axis movement mechanism.

A tool changer 20 is provided on the front side of the spindle head 7 and has a disk-shaped tool magazine 21 . The tool magazine 21 radially holds a plurality of tools (not shown) on its outer periphery. The tool changer 20 drives the tool magazine 21 by means of the magazine motor 55 (see FIG. 2), and positions the tool 4 designated by the tool change command at the tool change position. The tool changing position is the lowest position of the tool magazine 21 . The tool changer 20 replaces the tool 4 mounted on the spindle 9 and the tool attached to the tool magazine 21 by a series of operations of raising the spindle head 7, rotating the tool magazine 21, and lowering the spindle head 7. .

The control box 6 houses a numerical controller 30 (see FIG. 2). The numerical controller 30 controls a Z-axis motor 51, a spindle motor 52, an X-axis motor 53, and a Y-axis motor 54 (see FIG. 2) provided in the machine tool 1, and controls the table 13 and the tool 4 along the X-axis. direction, the Y-axis direction, and the Z-axis direction. In this case, the workpiece 3 fixed on the table 13 and the tool 4 mounted on the spindle 9 are moved relative to each other, and the workpiece 3 is subjected to various machining operations. The various types of processing include drilling using a drill, tap, or the like, and side surface processing using an end mill, milling cutter, or the like.

The operation panel 15 (see FIG. 2) is provided on the outer wall of a cover (not shown) that covers the machine tool 1, for example. The operation panel 15 includes an input section 16 and a display section 17 (see FIG. 2). The input unit 16 receives inputs such as various information and operation instructions, and outputs them to the numerical controller 30 . The display unit 17 displays various screens based on commands from the numerical controller 30 .

<Electrical configuration>
The electrical configuration of the numerical controller 30 and the machine tool 1 will be described with reference to FIG. The numerical control device 30 and the machine tool 1 are provided with a CPU 31, a ROM 32, a RAM 33, a storage device 34, an input/output section 35, drive circuits 51A to 55A, and the like. The CPU 31 centrally controls the numerical control device 30 . The ROM 32 stores main programs and the like. The main program is a program for executing main processing (see FIGS. 18 and 19). The main process reads the NC program line by line and executes various operations. The NC program consists of multiple lines containing various control commands including tool change commands. The control command of the NC program controls various operations, including axis movement of the tool 4 mounted on the base 2 and the spindle 9 of the machine tool 1, tool exchange, etc., line by line. The RAM 33 temporarily stores various information. The storage device 34 is non-volatile and stores NC programs and various information. The CPU 31 can store, in the storage device 34, an NC program that is input by the operator via the input unit 16 of the operation panel 15, as well as an NC program that is read from an external input.

The drive circuit 51A is connected to the Z-axis motor 51 and the encoder 51B. Drive circuit 52A is connected to spindle motor 52 and encoder 52B. The drive circuit 53A is connected to the X-axis motor 53 and the encoder 53B. Drive circuit 54A is connected to Y-axis motor 54 and encoder 54B. The drive circuit 55A is connected to the magazine motor 55 and the encoder 55B. The Z-axis motor 51, the main shaft motor 52, the X-axis motor 53, the Y-axis motor 54, and the magazine motor 55 are all servo motors. The drive circuits 51A-55A receive commands from the CPU 31 and output drive currents to the corresponding motors 51-55, respectively. The drive circuits 51A to 55A receive feedback signals from the encoders 51B to 55B and perform position and speed feedback control. The input/output unit 35 is connected to the input unit 16 and the display unit 17 of the operation panel 15, respectively. Hereinafter, when the motors 51 to 55 are collectively referred to as the motor 50 . When collectively referring to the drive circuits 51A to 55A, the drive circuit 50A is used.

<Outline of Correction Method for Command Trajectory>
In the numerical controller 30, it is necessary that the acceleration during movement of the base 2 and the tool 4 (hereinafter referred to as "controlled object") does not exceed the upper limit value that the machine tool 1 can handle. As an example of a method for preventing the upper limit value from being exceeded, there is a method in which the position command for moving the controlled object among the control commands of the NC program is smoothed by an acceleration/deceleration filter to limit the acceleration. When this method is used, portions of the movement trajectory of the controlled object that change steeply are smoothed. Generally, a multistage moving average filter is used as the acceleration/deceleration filter.

Note that the moving average filter has the characteristic of being able to remove specific frequency components. For this reason, a method of suppressing the vibration of the machine tool 1 by adjusting a moving average filter so as to remove the resonance frequency of the machine tool 1 has also been proposed.

On the other hand, when the position command is smoothed by the acceleration/deceleration filter, the movement trajectory according to the original position command (hereinafter referred to as the "command trajectory") after applying the acceleration/deceleration filter (referred to as "filter trajectory") may deviate, resulting in an error. As shown in FIG. 3A, for example, at the straight corner portions Cs and Ct of the commanded trajectory Gc(1), the corresponding filter trajectory Gf(1) is arranged inside (arrows Ys, Yt), or the commanded The radius of the corresponding filter locus Gf(1) may be smaller than the radius of the arc portion Cu of the locus Gc(1) (arrow Yu).

In response to this, numerical control device 30 corrects command locus Gc(1) in the following manner. First, the numerical controller 30 detects a plurality of points on the command trajectory Gc(1) (hereinafter each referred to as "base points") and points on the filter trajectory Gf(1) corresponding to the base points (hereinafter each referred to as "base points"). (referred to as "corresponding points"). Next, the numerical controller 30 calculates, for each base point, an error vector indicating the distance and direction from each base point to the corresponding corresponding point. Next, the numerical controller 30 calculates correction gains corresponding to the calculated error vectors for each base point.

Next, based on the error vector and the correction gain, the numerical controller 30 moves the base point in the direction in which the error between the command locus Gc(1) and the filter locus Gf(1) is canceled (arrows Zs, Zt , Zu, see FIG. 3B) to correct the command locus Gc(1). As shown in FIG. 3B, the corrected trajectory Gh(1), which is the corrected command trajectory Gc(1), is arranged on the opposite side of the filter trajectory Gf(1) from the command trajectory Gc(1). be. The moving direction of the correction trajectory Gh(1) with respect to the command trajectory Gc(1) is opposite to the error vector. The amount of movement of the correction trajectory Gh(1) with respect to the command trajectory Gc(1) is defined by the correction gain.

The numerical controller 30 performs filtering processing using an acceleration/deceleration filter for a movement command for moving the controlled object along the corrected locus Gh(1). In FIG. 3B, the trajectory after filtering is indicated as filtered trajectory Gf(1)'. The filter trajectory Gf(1)' corresponds to the movement trajectory along which the controlled object actually moves. In this case, the filter trajectory Gf(1)' is positioned slightly outside the command trajectory Gc(1) at positions before and after the straight corner portions Cs and Ct of the command trajectory Gc(1). However, the error between the commanded trajectory Gc(1) and the filtered trajectory Gf(1)' is significantly reduced compared to the error between the commanded trajectory Gc(1) and the filtered trajectory Gf(1). Furthermore, the filter trajectory Gf(1)' is at a position corresponding to the arc portion Cu of the command trajectory Gc(1), and the error from the command trajectory Gc(1) can be made substantially zero.

Note that the above method only corrects the position of the base point of the command locus Gc(1), and does not reduce the moving speed of the controlled object. Therefore, the error between the command locus Gc(1) and the filter locus Gf(1)' can be suppressed without increasing the cycle time during movement of the controlled object.

<How to determine corresponding points>
As described above, an error vector is required to correct the commanded trajectory Gc. The error vector indicates direction and magnitude. A specific method for calculating the error vector will be described below using the command locus Gc(2) shown in FIGS. 4 and 5 as an example. The command trajectory Gc(2) makes one round counterclockwise from the origin (0, 0). As the acceleration/deceleration filter, two stages of general moving average filters are used. In FIG. 5, the broken line indicates the filter locus Gf(2), which is the command locus Gc(2) after applying the acceleration/deceleration filter.

The error between the command trajectory Gc(2) and the filter trajectory Gf(2) is indicated by the distance between the base point on the command trajectory Gc(2) and the corresponding point on the corresponding filter trajectory Gf(2). be Therefore, how to determine the corresponding point on the filter locus Gf(2) corresponding to the base point on the command locus Gc(2) is important.

FIG. 5A shows corresponding points arranged on the filter trajectory Gf(2) in the range W1 of FIG. and are associated with each other. In this case, the direction of the error vector directed from the base point to the corresponding point abruptly changes before and after the straight corner portion Ca of the command locus Gc(2). Further, there is no reference point near the straight corner portion Ca, and no error vector exists. On the other hand, in order to correct so that the error from the command locus Gc(2) is small at the straight corner portion Ca, it is necessary to set an error vector also in the vicinity of the straight corner portion Ca of the command locus Gc(2). . Therefore, the error vector calculated by this method is inappropriate.

5(B) and (C) show, in the ranges W2 and W3 in FIG. 2) Shows the case where the above corresponding points are associated. In this case, as shown in FIG. 5B, in the range W2, the corner portion Cb of Gc(2) is associated with the corner portion of Gf(2), and the error vector from the base point to the corresponding point is appropriate. It can be seen that it is required for On the other hand, as shown in FIG. 5C, in range W3, there is a portion where the direction of the error vector from the base point to the corresponding point changes abruptly. The reason for this is that the points corresponding to the base point on the command locus Gc(2) are set at the portions before and after the corner portion Cb (see FIG. 5(B)) of the filter locus Gf(2). This is because Therefore, the error vector calculated by this method is also inappropriate.

In order to accurately correct the command trajectory Gc based on the error vector, it is necessary to appropriately calculate the corresponding points on the filter trajectory Gf corresponding to the base points on the command trajectory Gc. In this embodiment, corresponding points on the filter trajectory Gf are calculated by the following method.

A simple one-dimensional linear trajectory is assumed as the command trajectory Gc. In this case, the command locus Gc and the filter locus Gf match in shape. Therefore, the error between the command locus Gc and the filter locus Gf becomes zero. Also, the base point on the command locus Gc and the corresponding point on the filter locus Gf corresponding to this base point should coincide in position. Since a one-dimensional linear trajectory is assumed as the command trajectory Gc, matching the positions is equivalent to matching the distance from the starting point of the command trajectory Gc. In other words, in this method, the error is calculated with high accuracy by specifying corresponding points having the same distance from the starting point. Also, one corresponding point on the filter locus Gf corresponds to each base point on the command locus Gc.

The above concept is applied when the commanded trajectory Gc is a multi-dimensional trajectory. Here, as shown in FIG. 6, at the straight corner portion C of the command trajectory Gc, the filter trajectory Gf is arranged inside, and the distance of the filter trajectory Gf is shortened. Therefore, the distance from the starting point on the command locus Gc cannot be directly applied to the filter locus Gf. Therefore, in this method, a virtual linear trajectory that moves in a one-dimensional direction is assumed, considering only the movement in the tangential direction of the commanded trajectory Gc. As shown in FIG. 7, a one-dimensional filter trajectory Df is generated for this one-dimensional command trajectory Dc, and the distance from the starting point in the one-dimensional command trajectory Dc is directly applied to the one-dimensional filter trajectory Df.

As shown in FIG. 6, the base point p on the commanded trajectory Gc is determined by interpolating the portion between the start point and the end point of the commanded trajectory Gc. Interpolation is performed by determining, as base points p, the positions that a virtual point that starts moving from the starting point of the command locus Gc passes through in each interpolation cycle. In the command trajectory Gc, the i-th base point periodically determined from the starting point is denoted by _pi , and the tangential movement distance (distance from the starting point) at _pi is denoted by _si . Similarly, of the positions through which the virtual point that started moving from the starting point of the filter locus Gf passes in each interpolation cycle, the j-th point is denoted as _pj with circumflex.

On the other hand, as shown in FIG. 7, the i-th base point periodically determined from the starting point in the one-dimensional command trajectory Dc also has a moving distance of _si from the starting point. This is because the one-dimensional command trajectory Dc is a virtual linear trajectory that moves in a one-dimensional direction, considering only the movement in the tangential direction of the command trajectory Gc. is always equal. Similarly to the above, of the positions through which the virtual point that started moving from the starting point of the one-dimensional filter trajectory Df passes in each interpolation cycle, the j-th point is denoted as s _j with circumflex. .

As shown in FIGS. 8 and 9, the i- _{th si} on the command trajectory Gc or the one-dimensional command trajectory Dc is between the j-th ^s _j and the j+1-th ^s _j+1 on the one-dimensional filter trajectory Df. If it exists, it satisfies the relationship of the following formula (1-1).

なお、式（１－３）で示されるｕは、＾ｓ_ｊと＾ｓ_ｊ＋１との内分点に対応する係数（以降、「内分係数」と呼ぶ）である。つまり、指令軌跡Ｇｃのｉ番目の基点ｐ_ｉは、フィルタ軌跡Ｇｆのｊ＋ｕ番目の対応点＾（ｐ_ｊ＋ｕ）に対応し、その関係が式（１－２）で示されているということができる。 If the interpolation period is set to several milliseconds or less, there is no problem even if the moving speed is constant for each interpolation period. Therefore, when j that satisfies the relationship of formula (1-1) is used, the corresponding point ^(p _j+u ) of the filter trajectory Gf corresponding to the base point p _i of the command trajectory Gc is given by the following formula (1-2) It can be calculated as shown in (1-3).

Note that u in equation (1-3) is a coefficient (hereinafter referred to as an "internal division coefficient") corresponding to the point of internal division between ̂s _j and ̂s _j+1 . That is, it can be said that the i-th base point p _i of the command trajectory Gc corresponds to the j+u-th corresponding point ^(p _j+u ) of the filter trajectory Gf, and the relationship therebetween is expressed by equation (1-2). .

s _i of the command trajectory Gc or the one-dimensional command trajectory Dc, and ^s _j of the one-dimensional filter trajectory Df are given by equations (1-4) and (1-5) when a moving average filter is used as the acceleration/deceleration filter. can be shown. In the command trajectory Gc or the one-dimensional command trajectory Dc, the interpolation cycle is denoted as "T _s ", and the speed of the virtual point moving from the start point to the end point in the k-th interpolation cycle is denoted as "F _k ". , the time constant of the m-th acceleration/deceleration filter is expressed as "τ _m ", and the number of stages of the acceleration/deceleration filter is expressed as "M".

By substituting equations (1-1) to (1-3) into equations (1-4) and (1-5) and solving for j and _u , we obtain It is also possible to obtain the corresponding point ^(pj _+u ). However, in general, a plurality of acceleration/deceleration filters are used as shown in equations (1-4) and (1-5). In addition, there are many trajectories due to continuation of minute blocks, and the speed may change in a complicated manner. Therefore, it is difficult to solve for j and u analytically. Therefore, in this embodiment, the one-dimensional command trajectory and the one-dimensional filter trajectory are used to numerically search for corresponding points from equations (1-1) to (1-3).

FIG. 10 shows the error vector when the corresponding point ^(p _j+u ) for the base point p _i is determined based on equations (1-1) to (1-3). As shown in FIG. 10A, in the range W1, there is no abrupt change in the direction of the error vector before and after the straight corner portion Ca. A vector is wanted. Further, as shown in FIG. 10(C), in the range W3, the point corresponding to the base point of the command locus Gc(2) is closer to the corner portion Cb (see FIG. 10(B)) of the filter locus Gf(2). It is set in the later part and no abrupt change in the error vector is seen. Therefore, by determining the reference point p _i and the corresponding point ^(p _j+u ) based on the equations (1-1) to (1-3), it is possible to appropriately determine the error vector for correcting the commanded trajectory Gc. I understand.

<Correction for circular command locus>
Next, the error between the command trajectory Gc having a simple shape and the filtered trajectory Gf will be described. First, a simple arcuate command locus Gc is assumed. The radius of the commanded trajectory Gc is contracted by applying the acceleration/deceleration filter. The amount of contraction of the radius of the filter trajectory Gf with respect to the command trajectory Gc can be obtained from the frequency characteristics of the acceleration/deceleration filter. The X, Y coordinates of the arcuate command trajectory Gc can be described as motion of a single frequency component as shown in the following equation (2-1). Note that the radius of the arc is denoted as R, the angular velocity is denoted as ω, and the time is denoted as t.

Therefore, when the acceleration/deceleration filter is applied to (x, y), the X coordinate ^x and the Y coordinate ^y are simply influenced by the frequency characteristics (gain H(ω), phase Φ(ω)) of the acceleration/deceleration filter. , and changes as shown in the following formula (2-2).

Therefore, the contraction amount d _R of the arc radius is given by the following equation (2-3) regardless of the phase shift.

FIG. 11A shows a simple circular command locus Gc(3) and filter locus Gf(3). The error vector ε extending from each base point on the command trajectory Gc(3) toward the corresponding point on the corresponding filter trajectory Gf(3) is the center of the command trajectory Gc(3) and the filter trajectory Gf(3). It can be confirmed that it extends along the reference radial direction. That is, the magnitude of the error vector ε matches the contraction amount _dR of the radius of the filter trajectory Gf(3) with respect to the command trajectory Gc(3).

A parameter (hereinafter referred to as “correction vector”) −kε obtained by multiplying the correction gain k by the vector −ε pointing in the opposite direction to the error vector ε is defined. Then, based on the calculated correction vector -kε, the commanded trajectory Gc(3) is corrected so that the radius of the arc-shaped commanded trajectory Gc is increased. Then, the corrected trajectory Gh(3), which is the commanded trajectory Gc(3) after correction, is filtered by the acceleration/deceleration filter. As a result, the error between the original commanded trajectory Gc and the filtered trajectory Gf' after filtering the corrected trajectory Gh(3) can be reduced.

The error e between the filtered trajectory Gf(3)′ obtained by applying the acceleration/deceleration filter to the corrected trajectory Gh(3) obtained by correcting the commanded trajectory Gc(3) using the correction vector −kε and the original commanded trajectory Gc(3) is It is shown by the following formula (2-4). The angular velocity on the filter locus Gf(3)' is expressed as ω'.

Note that, as shown in FIG. 11A, the error vector ε is a vector extending along the radial direction and does not have a component in the circumferential direction. Therefore, the correction does not change the angular velocity and ω' is equal to ω. Therefore, equation (2-4) can be transformed into equation (2-5).

In actual motion, the arc radius R is greater than zero and the angular velocity ω is not zero. Therefore, the gain H(ω) of the acceleration/deceleration filter becomes smaller than one. Therefore, the error e can be reduced to 0 when the correction gain k(ω) satisfies the following equation (2-6).

FIG. 12 shows an example of the calculation result of formula (2-6). As analysis conditions, a two-stage (M=2) moving average filter was used as the acceleration/deceleration filter, and their time constants (τ ₁ and τ ₂ ) were set to 94 msec and 127 msec. FIG. 11B shows a corrected trajectory Gh(3) obtained by correcting the commanded trajectory Gc(3) with the correction vector −kε determined by the correction gain k derived from the result of FIG.

From the above results, it can be seen that in the case of the arc-shaped commanded trajectory Gc, the error between the commanded trajectory Gc and the filtered trajectory Gf' can be reduced to 0 by correcting the commanded trajectory Gc based on the correction vector -kε. . Further, it can be seen that the correction vector -kε can be easily obtained from the error e between the command locus Gc and the filter locus Gf and the angular velocity ω.

<Correction for straight corner portion C of commanded trajectory>
Next, the error between the commanded trajectory Gc and the filter trajectory Gf will be described with respect to the simple straight corner portion C of the commanded trajectory Gc. As shown in FIG. 13A, in the vicinity of the straight corner portion Cc of the command trajectory Gc(4), the filtered trajectory Gf(4) obtained by applying the acceleration/deceleration filter to the command trajectory Gc(4) is the command trajectory Gc(4 ) is placed inside. The reason for this is that applying the acceleration/deceleration filter to the command locus Gc(4) causes the points of the command locus Gc(4) within the range of the time constant of the acceleration/deceleration filter to influence each other. The time taken to pass through a portion of the filter trajectory Gf(4) located inside the command trajectory Gc(4) matches the time constant of the acceleration/deceleration filter.

FIG. 13B shows that the correction vector −kε used to suppress the error between the command locus Gc(4) and the filter locus Gf(4) is uniformly set to 1.5 in the vicinity of the straight corner portion Cc. A case of 5 (fold) is shown. Correction trajectory Gh(4) is arranged outside command trajectory Gc at straight corner portion Cc. FIG. 13C shows a filtered trajectory Gf(4)' obtained by applying the acceleration/deceleration filter to the corrected trajectory Gh(4). From this result, at the straight corner portion Cc of the command trajectory Gc(4), the error between the command trajectory Gc(4) and the filter trajectory Gf(4)' is the difference between the command trajectory Gc(4) and the filter trajectory Gf(4) It can be confirmed that the error is smaller than the error between

As described above, the points of the command locus Gc(4) within the range of the time constant of the acceleration/deceleration filter affect each other. Therefore, by correcting the command trajectory Gc(4) from the start point Pa to the end point Pb of the portion where the command trajectory Gc(4) and the filter trajectory Gf(4) are separated in the vicinity of the straight corner portion Cc, The error between the command locus Gc(4) and the filter locus Gf(4)' decreases in the vicinity of the straight corner portion Cc. Also, the filter locus Gf(4)' is located slightly outside the command locus Gc(4) in the vicinity of the start point Pa and the end point Pb. At Cc, the error from the filter locus Gf(4)' is greatly reduced.

Note that the error between the command locus Gc(4) and the filter locus Gf(4)' changes according to the magnitude of the correction gain k. However, unlike the case where the arc-shaped command trajectory Gc(3) (see FIG. 11) is assumed, in the straight corner portion Cc of the command trajectory Gc(4), the correction gain k that can make the error zero is not exist.

Therefore, in this embodiment, the area between the command locus Gc and the filter locus Gf is used in order to reduce the error in the entire vicinity of the straight corner portion C. FIG. Hereinafter, the area between the commanded trajectory Gc and the filtered trajectory Gf is referred to as "accumulated trajectory error". The accumulated trajectory error increases as the error between the command trajectory Gc and the filter trajectory Gf' placed outside near the start point Pa and the end point Pb of the portion where the trajectories are separated before and after the straight corner portion Cc increases. On the other hand, as the difference between the commanded trajectory Gc and the filtered trajectory Gf' decreases in the vicinity of the straight corner portion Cc, the accumulated trajectory error also decreases. Therefore, using the accumulated trajectory error to evaluate the effect of suppressing the error between the command trajectory Gc and the filter trajectory Gf' is suitable for the purpose of reducing the error in the vicinity of the straight corner portion C as a whole.

Here, it is assumed that the command locus Gc is corrected with a correction vector −kε having a constant correction gain k at each base point within the range of the angle formed by the straight corner portion C from 0 degrees to 180 degrees. FIG. 14 shows an example of the result of deriving the optimum correction gain k using the cumulative trajectory error as an evaluation function at each angle. As analysis conditions, a two-stage moving average filter was used as the acceleration/deceleration filter, and their time constants were set to 94 msec and 127 msec. As shown in FIG. 14, it can be seen that a correction gain k monotonically increasing with respect to the angle formed by the straight corner portion C is obtained. For this reason, in order to effectively suppress the error between the command locus Gc and the filter locus Gf' by calculating the correction gain k using the evaluation function in the command locus Gc shaped like the straight corner portion C. It becomes possible to appropriately determine the correction gain k of .

<Correction of Command Trajectory Having Arbitrary Shape>
Next, a method of calculating the correction gain k for correcting the command locus Gc having an arbitrary shape including an arc shape (see FIG. 11) and a straight corner portion C (see FIG. 13) will be described. The correction gain k is determined by making the error from the filter trajectory Gf zero in the arc-shaped portion of the command trajectory Gc and minimizing the accumulated trajectory error in the straight corner portion C. .

Assuming the command locus Gc(5) shown in FIG. 15A, the proposed method of calculating the correction gain k using the angle (direction) change at each base point p will be described. Note that the angular change at each base point p under a constant interpolation cycle is equivalent to the angular velocity when the controlled object moving along the command locus Gc(5) passes through the base point p. Therefore, hereinafter, the change in angle is referred to as "angular velocity".

As described with reference to FIGS. 11 and 13, it should be considered that the appropriate correction gain k changes based on the angular velocity in any case of the command trajectory Gc having an arbitrary shape including a circular arc and a straight corner portion C. can be done. This is because the change in the angle formed by the straight corner portion C corresponds to the change in angular velocity under constant speed. Since the angular velocity does not change over the entire circumference of the arcuate command locus Gc, the error magnitude |ε| and the optimum correction gain k remain unchanged over the entire circumference. On the other hand, at the straight corner portion C of the command trajectory Gc, the angular velocity changes in the shape of an impulse at the straight corner portion C, and the error is Since the magnitude |ε| changes, the optimum correction gain k also differs within this range.

A method of calculating the optimum correction gain k in the command locus Gc(5) based on the angular velocity ω using the correction gain function K(ω) described later will be described below. Further, in using the correction gain function K(ω), the angular velocity ω on the command locus Gc is filtered by an angular velocity filter _wl , which will be described later. Then, the correction gain k is derived by inputting the angular velocity ^ω filtered by the angular velocity filter _wl into the correction gain function K(ω). As described above, by using the angular velocity at each base point and performing the filtering process, it becomes possible to deal with the command trajectory Gc having an arbitrary shape with a single gain function.

<Correction gain function K(ω)>
In both the arc-shaped portion of the command locus Gc(3) shown in FIG. 11 and the straight corner portion Cc of the command locus Gc(4) shown in FIG. It increases gently and monotonically in the region where the angular velocity is low. Therefore, the correction gain function can be defined based on the relational expression in the arc-shaped command trajectory Gc shown in Equation (2-6). By using the corrected trajectory Gh(5) obtained by correcting the commanded trajectory Gc(5) using the correction gain k calculated by the correction gain function, the arc-shaped portion of the commanded trajectory Gc(5) is corrected to the commanded trajectory It is possible to suppress the error between Gc(5) and the filtered locus Gf(5)' obtained by applying the acceleration/deceleration filter to the corrected locus Gh(5).

In the relational expression (2-6), as shown in FIG. 12, the correction gain k diverges infinitely at a specific angular velocity. Therefore, in order to prevent an excessive load from being generated at the time of correction by the correction gain k, and also to correspond to an arbitrary trajectory other than a circular arc, a region including an angular velocity where the correction gain k diverges (hereinafter referred to as a "divergence region") ), it is necessary to replace the relational expression (2-6) with an approximation function that monotonically increases. On the other hand, in the low angular velocity region where no divergence occurs, by using the relational expression (2-6), the error between the arc-shaped command trajectory Gc(5) and the filter trajectory Gf(5)' is can be 0.

A case will be described below where an envelope function H _e (ω) is used as the approximation function in the divergence region. The envelope function H _e (ω) is a function when the sine function of the denominator of the gain H(ω), which is the frequency characteristic of the acceleration/deceleration filter, is set to 1. The correction gain k _e (ω) when using the envelope function H _e (ω) satisfies the relational expression (3-1).

When the correction gain k(ω) becomes discontinuous, it is expected that the correction trajectory Gh(5), which is the command trajectory Gc(5) after correction, will change abruptly, causing an increase in acceleration. For this reason, two relational expressions (2-6) and (3-1) are used so that the correction gain k(ω) changes continuously, and the correction gain function K(ω) is changed to the following expression (3 -2).

補正項ｋ_ｔは、以下の式（３－４）の関係を満たす値である。

補正ゲイン関数Ｋ（ω）をグラフで表したものを、図１６に示す。 In equation (3-2), ±ω _t is the angular velocity that switches between the two functions. _kt is an offset amount for continuously connecting two functions. Since the correction gain function K(ω) changes smoothly at the angular velocity at which the two functions are switched, the angular velocity at which the slopes of the respective functions become equal is defined as the angular velocity ± _ωt . The angular velocity ±ω _t that satisfies this condition is a value that satisfies equation (3-3).

The correction term k _t is a value that satisfies the relationship of Equation (3-4) below.

A graphical representation of the correction gain function K(ω) is shown in FIG.

<Angular velocity filter>
As noted above, the correction gain function K(ω) is calculated as a function of angular velocity ω. On the other hand, in the case of the straight corner portion C, the angular velocity ω changes in an impulse-like manner only at the straight corner portion C, and becomes zero at the other straight portions. However, the trajectory error does not occur only in the straight corner portion C, but occurs over the time constant range (range of influence) of the acceleration/deceleration filter. It is preferable to perform not only C but, for example, from the start point Pa to the end point Pb shown in FIG. 13C (the time constant range of the acceleration/deceleration filter). Therefore, the command trajectory Gc is corrected based on the angular velocity (instantaneous value) at one point of the straight corner portion C so that the correction by the correction gain function K(ω) can be performed in the range from the start point Pa to the end point Pb. It is necessary to

The angular velocity filter _wl is used to smooth changes in the angular velocity ω of the commanded trajectory Gc when the correction gain function K(ω) is applied to an arbitrary commanded trajectory Gc. A solid line L(1) in FIG. 15(B) indicates the temporal change of the angular velocity ω on the command locus Gc(5). A broken line L(2) indicates the temporal change of the angular velocity ^ω after the filtering process is executed by the angular velocity filter _wl . The two impulse-like angular velocity changes shown in the figure occur at the straight corner portions p(1) and p(2), and the angular velocity is a constant value −ω _f It has become. That is, the angular velocity changes at the straight corner portion and the circular arc portion show extreme characteristics of a steepest impulse-like change and a constant value with the least change. Therefore, by setting the filtering processing of the angular velocity and the correction gain function so as to obtain an appropriate correction gain for both the straight corner portion and the circular arc portion, an appropriate correction gain can be obtained for an arbitrary trajectory. demand.

First, in the arc-shaped portion of the command locus Gc(3) shown in FIG. 11 and the straight corner portion Cc of the command locus Gc(4) shown in FIG. and the correction gain k(ω) monotonously increases with an increase in the angular velocity ω, except for a region where the angular velocity ω is extremely large. Therefore, the correction gain function K(ω) can be extended to any trajectory other than the arcuate command trajectory Gc by performing filtering processing that satisfies the following characteristics on the command trajectory Gc.
(1) When the change in the angular velocity ω is steep, filtering processing is performed to reduce the steep change in the angular velocity ω.
(2) When the change in the angular velocity ω is steep, the filtering process expands the influence of the change in the angular velocity ω to a predetermined range around the time when the sharp change in the angular velocity ω occurs.

On the other hand, in the case of the arcuate portion of the command locus Gc, it is necessary to avoid changes in the angular velocity ω even when the filtering process is performed. Therefore, in addition to (1) and (2), filtering processing that satisfies the following characteristics must be performed on the command trajectory Gc. (3) When the change in the angular velocity ω is moderate, the filtering process prevents the angular velocity ω from changing greatly. In particular, when a constant angular velocity ω continues over a predetermined range (influenced time constant range) like an arc, the value of the angular velocity ω does not change at all due to the filtering process.

As filtering processing satisfying the conditions (1) to (3), filtering processing is performed using a low-pass filter having the same time constant as the acceleration/deceleration filter. Hereinafter, a low-pass filter used for filtering the commanded trajectory Gc is referred to as an "angular velocity filter _wl ". The angular velocity filter _wl applied here is an example, and its weighting function W(t) is given by the following equation (4-1). W _e , β _e , W _c , and β _c are parameters for determining the shape of the filter, respectively.

Moreover, the above parameters satisfy the following equation (4-2) so as not to change the original angular change amount (integral value of angular velocity).

FIG. 17 shows an example of the weighting function W(t) calculated based on Equation (4-2). The shape of the weighting function W(t) is determined by the parameters W _e , β _e , W _c , β _c . The shape of the weighting function W(t) can be freely adjusted from a shape with constant coefficients like a moving average filter to a shape with large coefficients only near the middle of the filter (changes in coefficients are large). By adjusting the shape of the weighting function W(t), it is possible to adjust how much the steep changes in the angular velocity ω are reduced and spread out over the filter's time constant range. Adjustment of the parameters W _e , β _e , W _c , β _c is performed by numerical calculation. More specifically, the parameters W _e , β _e , W _c , and β _c that minimize the average value of the accumulated trajectory errors when the angle formed by the straight corner portion C is changed from 0 degrees to 180 degrees. may be adjusted by numerically searching for

Here, _τt is equal to the time constant of the acceleration/deceleration filter. Equal time constant means equal to the total time constant obtained by adding the time constants of all the filters when the acceleration/deceleration filter is composed of a plurality of filters.

After application of the angular velocity filter _wl , the angular velocity ω on the command locus Gc is input as an input variable of the correction gain function K(ω). Thereby, the correction gain k is finally calculated.

<Main processing>
The main processing will be described with reference to FIGS. 18 and 19. FIG. The main process is started by the CPU 31 reading out and executing the NC program stored in the ROM 32 when an instruction to start machining based on the NC program is input via the input unit 16 .

As shown in FIG. 18, the CPU 31 initializes variables i and j stored in the RAM 33 by setting them to 0 (S11). The CPU 31 sequentially reads one line of the position command from the NC program stored in the storage device 34 as necessary (when the previous interpolation is completed) (S13). The read position command defines the position of the controlled object after movement. The CPU 31 executes interpolation processing in advance based on the read position command every interpolation cycle, which is a predetermined processing cycle (S15). In the interpolation processing, the commanded position of the commanded trajectory Gc extending from the position of the controlled object before movement to the position of the controlled object after movement is interpolated with a plurality of base points p _j . A plurality of reference points p _j are points that the controlled object passes through in each interpolation cycle when it moves at the command speed from the position before movement to the position after movement. Next, the CPU 31 calculates the tangential movement distance _sj and the angular velocity _ωj at the reference point _pj of the command locus Gc (S17).

The CPU 31 performs a filtering process using an acceleration/deceleration filter on the base points up to _pj interpolated by the interpolation process, and calculates the j-th point ^ _pj on the filter locus Gf (S19).

The CPU 31 performs a filtering process using an acceleration/deceleration filter on the virtual one-dimensional linear trajectory having the calculated movement distance _sj , and calculates ^ _sj (S21). Further, the CPU 31 executes filtering processing by the angular velocity filter w _l (see formula (4-2)) on the calculated angular velocity up to ω _j to calculate ^ω _j (S23).

The CPU 31 determines whether or not there is a movement distance _si that satisfies the formula (1-1) (S25). When the CPU 31 determines that there is no movement distance s _i that satisfies the formula (1-1) (S25: NO), the process proceeds to S49 (see FIG. 19). As shown in FIG. 19, the CPU 31 adds 1 to the variable j and updates it (S49). The CPU 31 reads out all the position commands of the NC program stored in the storage device 34 by the processing of S13 (see FIG. 18), and determines whether or not the correction of the final base point after the interpolation has been completed (S51). When the CPU 31 determines that there remains a base point for which error correction has not been completed (S51: NO), the process returns to S13 (see FIG. 18). As shown in FIG. 18, the CPU 31 repeats the processes of S13 to S23 based on the updated variable j.

When the CPU 31 determines that there is a moving distance _si that satisfies the formula (1-1) (S25: YES), the process proceeds to S27. The CPU 31 substitutes s _i , ̂s _j , and ̂s _j+1 into equation (1-3), and calculates an internal division coefficient u between ̂s _j and ̂s _j+1 (S27). The CPU 31 advances the process to S31.

As shown in FIG. 19, the CPU 31 substitutes the internal division coefficient u calculated by the processing of S27 (see FIG. 18) into the equation (1-2) _to A corresponding point ̂p _j+u above is determined (S31). The CPU 31 calculates an error vector ε _i indicating the direction and distance from the base point p _i to the corresponding point ^p _{j + u} based on the following equation (5-1) (S33).

The CPU 31 substitutes the internal division coefficient u calculated by the processing of S27 (see FIG. 18) into the following equation (5-2) to calculate the angular velocity ^ω _j+u at the corresponding point ^p _j+u on the filter trajectory Gf ( S35).

The CPU 31 substitutes the calculated angular velocity ^ω _j+u into the equation (3-2) to calculate the correction gain k _i (S37). The CPU 31 calculates a correction vector −k _{i ε i} by multiplying the calculated correction gain k i by the error vector _{ε i} calculated in the process of S33 and reversing the direction (S39).

The CPU 31 calculates a point q _i by moving the position of the reference point p _i on the command locus Gc based on the correction vector −k _i ε _i calculated in the process of S39 (S41). Note that the trajectory passing through the point _qi corresponds to the corrected trajectory Gh. Therefore, the commanded trajectory Gc is corrected to the corrected trajectory Gh by the above processing.

The CPU 31 performs a filtering process using an acceleration/deceleration filter on the calculated points up to _qi (S43). The CPU 31 drives the controlled object based on the position command indicating the position of the point ^ _qi after filtering (S45). The CPU 31 updates the variable i by adding 1 (S47). The CPU 31 updates the variable j by adding 1 (S49). The CPU 31 determines whether the correction and control processing has been completed up to the last base point (S51). When the CPU 31 determines that there are base points that have not been processed (S51: NO), the process returns to S13 (see FIG. 18).

When the CPU 31 determines in the process of S13 (see FIG. 18) that the process has been completed up to the last base point (S51: YES), it ends the main process.

<Evaluation results>
Performance evaluation of the machine tool 1 driven as described above was performed by experiment. The output signals of the encoders 53B and 54B attached to the X-axis motor 53 and the Y-axis motor 54 that drive the base 2 of the machine tool 1 are recorded, and the position and speed of the table 13 and the filter trajectory Gf with respect to the command trajectory Gc, The error of Gf' was evaluated. Further, as shown in FIG. 20, a beam vibrating in the X-axis direction and a beam vibrating in the Y-axis direction are provided on the table 13 of the machine tool 1, and an accelerometer is attached to each beam to vibrate in the X-axis direction and the Y-axis direction. Directional vibration was measured, and the vibration suppression effect was also evaluated. FIG. 21 shows a co-advance frequency of 10.6 Hz detected by an accelerometer attached to a beam vibrating in the X-axis direction and a co-advance frequency of 7.0 Hz detected by an accelerometer attached to a beam vibrating in the Y-axis direction. Vibration at 86 Hz is shown.

A two-stage acceleration/deceleration filter was used in the evaluation. The time constants of the two-stage acceleration/deceleration filters are common to the case of application to the command locus Gc and the case of application to the correction locus Gh. The time constants were set to 94 (≈1000/10.6) msec and 127 (≈1000/7.86) msec so as to suppress the vibration of the resonance frequencies of the two beams. As the angular velocity filter, a low-pass filter having a weighting function having a shape shown in FIG. 22 was obtained by numerically searching for a parameter that minimizes the cumulative trajectory error. The speed of the controlled object (base 2) in the position command was set to 6000 mm/min. FIG. 23 shows the measurement results of the cumulative trajectory error.

<Evaluation result 1>
A command locus Gc(6) having a straight corner portion Cd was evaluated. FIG. 24 shows a command trajectory Gc(6), a filtered trajectory Gf(6) obtained by applying an acceleration/deceleration filter to the command trajectory Gc(6), and a corrected trajectory Gh obtained by correcting the command trajectory Gc(6) applying an acceleration/deceleration filter. The applied filter locus Gf(6)' is shown. FIG. 25 shows respective errors of the filter trajectories Gf(6) and Gf(6)' with respect to the command trajectory Gc(6). FIG. 26 shows the velocity of the controlled object moving along the filter loci Gf(6) and Gf(6)', respectively. FIG. 27 shows the analysis result of the vibration frequency of the beam.

From the results shown in FIG. 24(A), it was found that the filter locus Gf(6)' near the straight corner portion Cd is arranged outside the command locus Gc(6). Also, from the results shown in FIG. 25, it has been clarified that the error between the commanded trajectory Gc(6) and the filtered trajectory Gf(6)' is smaller than that of the filtered trajectory Gf(6). Further, from the results shown in FIG. 23, it was confirmed that the accumulated trajectory error of 21.65 mm ² in the filter trajectory Gf(6) decreased to 11.37 mm ² in the filter trajectory Gf(6)'. Further, from the results shown in FIG. 26, although there is a portion where the speed slightly exceeds 6000 mm/min in the filter locus Gf(6)', the time when the tangential direction speed becomes 0, that is, the timing at which the movement is completed is before and after the correction. , and the cycle time did not increase with the correction. Further, from the results shown in FIG. 27, it was found that both the filter locus Gf(6) and the filter locus Gf(6)' had a similarly small vibration component and a sufficient vibration suppression effect.

From the above, by correcting the command trajectory Gc(6) and deriving the filter trajectory Gf(6)' by the method described in this embodiment, the effect of suppressing vibration is not impaired and the cycle time is not changed. In addition, it has been found that it is possible to reduce the error from the command locus Gc(6).

<Evaluation result 2>
The command trajectory Gc(7) including an arc-shaped portion was evaluated. FIG. 28 shows a command trajectory Gc(7), a filtered trajectory Gf(7) obtained by applying an acceleration/deceleration filter to the command trajectory Gc(7), and a corrected trajectory Gh(7) obtained by correcting the command trajectory Gc(7). A filter locus Gf(7)' to which the deceleration filter is applied is shown. FIG. 29 shows the error of each of the filter trajectories Gf(7) and Gf(7)' with respect to the command trajectory Gc(7). FIG. 30 shows the velocity of the controlled object moving along the filter loci Gf(7) and Gf(7)', respectively.

From the results shown in FIGS. 28 and 29, it is clear that the error between the commanded trajectory Gc(7) and the filtered trajectory Gf(7)' is smaller than that of the filtered trajectory Gf(7). In particular, it was confirmed that the error between the command trajectory Gc(7) and the filter trajectory Gf(7)' was almost zero in the arcuate range W7 of the command trajectory Gc(7). Also, from the results shown in FIG. 23, it was confirmed that the cumulative trajectory error was reduced to 20.2% over the entire trajectory. Further, from the results shown in FIG. 30, although there are places where the velocity of the filter locus Gf(7)' slightly exceeds the velocity of the command locus Gc(7), the velocity of the filter locus Gf(7) before correction is It was found that there was almost no change in

From the above, by correcting the command trajectory Gc(7) and deriving the filter trajectory Gf(7)' by the method described in this embodiment, the error between the command trajectory Gc(7) and the command trajectory Gc(7) can be greatly reduced. It was found that the error can be reduced to almost 0, especially in the arc-shaped range.

<Evaluation result 3>
A complex command trajectory Gc(8) composed of a series of linear trajectories of minute length was evaluated. The command trajectory Gc(8) consists of only a short straight line with a length of 1 mm and goes around counterclockwise from coordinates (0, 0). FIG. 31 shows a command trajectory Gc(8), a filtered trajectory Gf(8) obtained by applying an acceleration/deceleration filter to the command trajectory Gc(8), and a corrected trajectory Gh(8) obtained by correcting the command trajectory Gc(8). A filter locus Gf(8)' to which the deceleration filter is applied is shown. FIG. 32 shows the error of each of the filter trajectories Gf(8) and Gf(8)' with respect to the command trajectory Gc(8). FIG. 33 shows the velocity of the controlled object moving along the filter loci Gf(8) and Gf(8)', respectively.

From the results shown in FIGS. 31 and 32, it is clear that the error between the commanded trajectory Gc(8) and the filtered trajectory Gf(8)' is smaller than that of the filtered trajectory Gf(8). Also, from the results shown in FIG. 23, it was confirmed that the cumulative trajectory error was reduced to 44.8% over the entire trajectory. Further, from the results shown in FIG. 33, the velocity of the filter locus Gf(8)' only slightly increases with respect to the velocity of the command locus Gc(8). It was found that the overall speed was close to the original command speed of 6000 m/min compared to the speed of .

From the above, by correcting the command trajectory Gc(8) and deriving the filter trajectory Gf(8)′ by the method described in this embodiment, the command trajectory Gc(8) can be corrected without changing the cycle time. It was found that it is possible to reduce the error between

<Actions and effects of the present embodiment>
The numerical controller 30 calculates the error vector ε and the correction gain k for each base point p on the command locus Gc (S33, S37). The numerical controller 30 applies the calculated error vector ε and the correction gain k to each of the reference points p to correct the commanded trajectory Gc (S41). The numerical controller 30 filters the corrected trajectory Gh, which is the corrected command trajectory Gc, with an acceleration/deceleration filter to calculate a filtered trajectory Gf', and controls the controlled object based on the filtered trajectory Gf'. In this case, the numerical controller 30 can appropriately approximate the command trajectory Gc and the filter trajectory Gf' according to the angular velocity determined by the shape of the original command trajectory Gc and the movement speed. Therefore, when the command trajectory Gc is filtered by the acceleration/deceleration filter to control the controlled object, the numerical controller 30 corrects the error between the actual movement trajectory (filtered trajectory Gf′) of the controlled object and the original command trajectory Gc. It can be suppressed with high accuracy.

The numerical controller 30 interpolates a plurality of base points p on the command locus Gc extending between two positions defined by the position command (S15). Therefore, the numerical control device 30 can correct the command trajectory Gc using more points, so the error between the filter trajectory Gf' corresponding to the actual movement trajectory of the controlled object and the original command trajectory Gc is can be suppressed with high accuracy.

The numerical controller 30 calculates the tangential movement distance s at the reference point p of the commanded trajectory Gc (S17). The numerical control device 30 determines the base point p based on the calculated moving distance s and ^s, which is the result of filtering the virtual one-dimensional linear trajectory having the moving distance s using an acceleration/deceleration filter. A corresponding point ^p on the filter locus Gf is determined (S31). In this case, the numerical controller 30 can accurately determine the corresponding points ^p as the points on the filter locus Gf corresponding to the base points p on the command locus Gc.

The numerical controller 30 calculates the angular velocity ω in the tangential direction at the reference point p of the command locus Gc (S17). A correction gain k is calculated based on the calculated angular velocity ω (S37). In this case, the numerical controller 30 can calculate the optimum correction gain k for each base point p on the command locus Gc based on the angular velocity ω.

The numerical controller 30 calculates the correction gain k by substituting the angular velocity ω into the equation (3-2) (S37). Here, Equation (3-2) is derived by applying the envelope function H _e in a region (divergence region) where the angular velocity ω is greater than ω _t . That is, the numerical controller 30 calculates a correction gain k that allows the filter trajectory Gf to match the arc-shaped command trajectory Gc from the command trajectory Gc to the envelope curve (see FIG. 16) of the graph plotted for each angular velocity. The correction gain k is calculated by applying the calculated angular velocity ω. Therefore, the numerical controller 30 can calculate the optimum correction gain k for each base point p on the command locus Gc according to the value of the angular velocity ω.

The numerical controller 30 filters the calculated angular velocity ω using an angular velocity filter having the same time constant as the acceleration/deceleration filter (S23). The numerical controller 30 calculates a correction gain k based on the filtered angular velocity (S37). Thereby, the numerical controller 30 can calculate the optimum correction gain k even when the instantaneous change in the angular velocity ω is large.

<Modification>
The present invention is not limited to the above embodiments, and various modifications are possible. The main process described above is not limited to being executed by the numerical controller 30 , and may be executed by the machine tool 1 or by a PC connected to the machine tool 1 .

The filter used when the numerical controller 30 performs the filtering process in S19 and S21 is not limited to the moving average filter, and other known filters may be used. The numerical control device 30 may specify corresponding points using the position defined by the position command as a base point without performing the interpolation processing in S15.

The method of determining corresponding points in S31 is not limited to the above embodiment, and other methods may be used. For example, the numerical controller 30 analytically obtains corresponding points after filtering by an acceleration/deceleration filter for the coordinate information indicating the position of the base point p (for example, when a one-stage moving average filter is used as the acceleration/deceleration filter, the formula ( 1-4) It is possible to substitute equations (1-1) to (1-3) into (1-5) and analytically solve for j and u), and determine corresponding points based on the result You may

The method of calculating the correction gain in S37 is not limited to the above embodiment, and other methods may be used. For example, the numerical control device 30 may store in advance a correction gain function determined according to the shape and speed of the command trajectory Gc for each command trajectory Gc, and determine the correction gain by using the corresponding correction gain function. good. In this case, the angular velocity ω may not be used when determining the correction gain.

The numerical controller 30 calculated the correction gain using the correction gain function K(ω) derived by applying the envelope function _He . On the other hand, the numerical controller 30 calculates the correction gain using a method other than applying the envelope function _He , for example, a method such as setting a maximum value so that the correction gain function K(ω) does not diverge. You may

The numerical controller 30 filtered the angular velocity ω with an angular velocity filter in S23, and calculated the correction gain based on the result. On the other hand, the numerical control device 30 does not use the angular velocity filter, for example, obtains the angular velocity that changes along the filter locus ^p after applying the acceleration/deceleration filter to the command locus, and uses it to calculate the correction gain. You may
At S45, the speed may exceed a predetermined limit speed. FIGS. 34 and 35 are flow charts of modifications to prevent the speed limit from being exceeded. 18 and 19 denote the same processing, so description thereof will be omitted.
The CPU 31 sets variables i, j, k, and l stored in the RAM 33 to 0, and also initializes the variables and the look-ahead list by emptying the look-ahead list, which will be described later (S111). i is the target starting point in the main process, j is the starting point during the prefetching, k is the index of the command block c during the prefetching of the NC program, and l is the counter. The first index is 0. The CPU 31 reads the command block c _k from the NC program stored in the storage device 34 . The NC program contains commanded positions and commanded velocities. The CPU 31 adds the command block _ck to the look-ahead list and stores it in the RAM 33 (S113). The CPU 31 executes interpolation processing for all command blocks stored in the look-ahead list stored in the RAM 33 (S115). In the interpolation processing, the command position on the command trajectory Gc extending from the position of the controlled object before movement to the position of the controlled object after movement is interpolated using a plurality of base points p _j . A plurality of reference points p _j are points that the controlled object passes through every predetermined interpolation cycle when it moves at the command speed from the position before movement to the position after movement. The CPU 31 stores each reference point p _j in the RAM 33 in association with the index of the instruction block before interpolation. Next, the CPU 31 executes S17 to S23 and determines whether or not the movement distance ^sj ₊₁ that satisfies ^ _sj ≤ _si+f/2 ≤ ^ _sj+1 has been calculated (S125). f, which represents the time constant of the filter by the number of base points, is calculated by the sum of time constants τt/control period=f. That is, the filtering process requires f base points. If ^s _j+1 has been calculated (S125: YES), the CPU 31 substitutes if/2 for the variable l (S127), and shifts the process to S128. This is because the look-ahead list of command blocks stores as many command blocks as are necessary for calculating the correction vector k _i+f/2 ε _i+f/2 . When the CPU 31 determines that the moving distance ̂s _j+1 has not been calculated (S125: NO), it adds 1 to the variable k (S126) and returns the process to S113. In S128, the CPU 31 substitutes s _i , ̂s _j , and ̂s _j+1 into equation (1-3), and calculates an internal division coefficient u between ̂s _j and ̂s _j+1 . However, variable i in equation (1-3) is changed to variable l. Then, the internal division coefficient u is substituted into the equation (1-2) to determine the corresponding point ^ _pj+u on the filter locus Gf corresponding to the base point p _l on the command locus Gc (S132). The CPU 31 calculates an error vector ε _l indicating the direction and distance from the reference point p _l to the corresponding point ^p _{j + u} based on the equation (5-1) (S133). The CPU 31 substitutes the internal division coefficient u into the equation (5-2) to calculate the angular velocity ̂ω j _+u at the corresponding point ̂p _j+u on the filter locus Gf (S135). The CPU 31 substitutes the calculated angular velocity ^ω _j+u into the equation (3-2) to calculate the correction gain _kl (S137). The CPU 31 multiplies the calculated correction gain k _l by the error vector ε _l calculated in the process of S133 to reverse the direction, thereby calculating a correction vector −k _l ε _l (S139). The CPU 31 calculates a point q _l by moving the position of the base point p _l on the command locus Gc based on the correction vector −k _l ε _l calculated in the process of S139 (S141).
After executing S141, the CPU 31 determines whether or not the variable l becomes i+f/2 (S142). When the variable l=i+f/2 (S142: YES), the CPU 31 advances the process to S144. When the variable l does not become i+f/2 (S142: NO), the CPU 31 adds 1 to the variable l and returns the process to S128. The CPU 31 performs a filtering process using an acceleration/deceleration filter on the calculated points from _qif/2 to qi _+f/2 to calculate ^ _qi (S144). Thereafter, the CPU 31 determines whether or not i=0 (S145), and if i=0 (S145: YES), the process proceeds to S149, which will be described later. When the CPU 31 determines that i is not 0 (S145: NO), the speed of _qi is calculated from the difference between the command block indicating the position of the base point _qi after filtering and the command block of qi _-1. (S146), it is determined whether the calculated speed of q _i exceeds the command speed of the command block _cm (m is the index of the command block) stored in association with p _if/2 (S147). When the CPU 31 determines that the speed of ^ _qi exceeds the command speed (S147: YES), it calculates "command speed of command block _cm /speed of ^ _qi " and uses the calculated result as the command speed. (S148), and the process proceeds to S115. Therefore, all command blocks stored in the look-ahead list are at or below the command speed. When determining that the speed of ^q _i does not exceed the command speed (S147: NO), the CPU 31 drives the controlled object based on the command block indicating the position of the point ^ _qi after filtering. Regarding the command block _cm stored in association with p _i-f , if the command block _cm-2 is included in the look-ahead list of command blocks, the CPU 31 deletes the command block (S149). The CPU 31 determines whether the correction and control processing has been completed up to the last base point (S151). When the CPU 31 determines that there remains a base point that has not been processed (S151: NO), the process returns to S113. When the CPU 31 determines that the process has been completed up to the last base point (S151: YES), the main process is terminated. Numerical controller 30 may use a similar method to correct acceleration so that it does not exceed the commanded acceleration when the acceleration exceeds the original commanded acceleration.

<Others>
CPU31 which processes S13 is an example of the "acquisition part" of the present invention. The acceleration/deceleration filter is an example of the "first filter" of the present invention. CPU31 which processes S19 and S21 is an example of the "first filtering part" of the present invention. CPU31 which processes S31 is an example of the "decision part" of the present invention. CPU31 which processes S33 is an example of the "vector calculation part" of the present invention. CPU31 which processes S37 is an example of the "gain calculation part" of the present invention. CPU31 which processes S41 is an example of the "trajectory correction part" of the present invention. CPU31 which processes S15 is an example of the "interpolation part" of the present invention. The CPU 31 that performs the process of S17 is an example of the "distance calculator" and "angular velocity calculator" of the present invention. The angular velocity filter is an example of the "second filter" of the present invention. CPU31 which processes S23 is an example of the "second filtering part" of the present invention. The process of S13 is an example of the "acquisition step" of the present invention. The processing of S21 is an example of the "first filtering step" of the present invention. The process of S31 is an example of the "determining step" of the present invention. The processing of S33 is an example of the "vector calculation step" of the present invention. The processing of S37 is an example of the "gain calculation step" of the present invention. The processing of S41 is an example of the "trajectory correction step" of the present invention. CPU31 which processes S144 is an example of the "third filtering part" of the present invention. CPU31 which processes S146 is an example of the "speed calculation part" of the present invention. S147
is an example of the "speed determination section" of the present invention. CPU31 which processes S148 is an example of the "speed change part" of the present invention.

1: machine tool 4: tool 13: table 30: numerical controller 31: CPU

Claims (11)

an acquisition unit that acquires a command for moving a controlled object that is at least one of a machine tool table and a tool;
a first filtering unit that applies a first filter to a command trajectory, which is a movement trajectory when the controlled object moves according to the command acquired by the acquisition unit;
a determination unit that determines a corresponding point on the filtered trajectory, which is a result of applying the first filter to the commanded trajectory by the first filtering unit, as a point corresponding to a base point that is a point on the commanded trajectory;
a vector calculation unit for calculating, for each base point, an error vector indicating the distance and direction from each of the base points to each of the corresponding corresponding points determined by the determination unit;
a gain calculation unit that calculates, for each base point, a gain corresponding to the error vector calculated by the vector calculation unit;
a trajectory correction unit that corrects the command trajectory by applying the error vector calculated by the vector calculation unit and the gain calculated by the gain calculation unit to the corresponding base points;
A numerical control device comprising: the command defines at least two positions of the controlled object;
A point that passes through each control cycle of the machine tool when the controlled object moves between the at least two positions defined by the command acquired by the acquiring unit at the command speed acquired by the acquiring unit, Further comprising an interpolation unit that interpolates as the base point,
The first filtering unit is
2. The numerical controller according to claim 1, wherein said first filter is applied to said command locus passing through said base point interpolated by said interpolator. further comprising a distance calculation unit for calculating a movement distance in a tangential direction at the reference point interpolated by the interpolation unit based on the commanded trajectory;
The first filtering unit is
Applying the first filter to a one-dimensional trajectory based on the movement distance calculated by the distance calculation unit,
The decision unit
3. The numerical controller according to claim 2, wherein the one-dimensional trajectory based on the movement distance to which the first filter is applied is used to determine the corresponding point corresponding to the base point. further comprising an angular velocity calculator that calculates an angular velocity at the reference point interpolated by the interpolator based on the commanded trajectory;
The gain calculation unit
3. The numerical controller according to claim 2, wherein the gain is calculated based on the angular velocity calculated by the angular velocity calculator. The gain calculation unit
A function defined based on the relationship between the gain and the angular velocity that allows the filter trajectory to match the command trajectory when the command trajectory is circular is applied to the angular velocity calculated by the angular velocity calculator. 5. The numerical controller according to claim 4, wherein the gain is calculated for each of the angular velocities by applying. The function is
6. A numerical controller according to claim 5, wherein said gain indicates an envelope of a graph plotted for each angular velocity. A second filtering unit that applies a second filter having the same time constant as the first filter to the angular velocity calculated by the angular velocity calculating unit,
The gain calculation unit
7. The numerical controller according to claim 4, wherein the gain is calculated based on the angular velocity after applying the second filter by the second filtering unit. a third filtering unit that applies the first filter to points on the corrected trajectory corrected by the trajectory correcting unit;
a speed calculation unit that calculates the speed of the point corresponding to the base point after filtering processing by the third filtering unit;
a speed determination unit that determines whether the speed calculated by the speed calculation unit exceeds the command speed;
and a speed change unit for changing the speed calculated by the speed calculation unit to be equal to or less than the command speed when the speed determination unit determines that the speed calculated by the speed calculation unit exceeds the command speed. 7. A numerical controller according to any one of claims 2 to 6. a third filtering unit that applies the first filter to points on the corrected trajectory corrected by the trajectory correcting unit;
a speed calculation unit that calculates the speed of the point corresponding to the base point after filtering processing by the third filtering unit;
a speed determination unit that determines whether the speed calculated by the speed calculation unit exceeds the command speed;
and a speed change unit for changing the speed calculated by the speed calculation unit to be equal to or less than the command speed when the speed determination unit determines that the speed calculated by the speed calculation unit exceeds the command speed. 8. A numerical controller according to claim 7. an acquiring step of acquiring a command for moving a controlled object, which is at least one of a machine tool table and a tool;
a first filtering step of applying a first filter to a command trajectory, which is a movement trajectory when the controlled object moves according to the command acquired by the acquiring step;
a determining step of determining a corresponding point on the filtered trajectory, which is the result of applying the first filter to the commanded trajectory in the first filtering step, as a point corresponding to a base point that is a point on the commanded trajectory;
a vector calculation step of calculating, for each of the base points, an error vector indicating the distance and direction from each of the base points to each of the corresponding corresponding points;
a gain calculation step of calculating, for each base point, a gain corresponding to the error vector calculated by the vector calculation step;
a trajectory correction step of correcting the command trajectory by applying the error vector calculated by the vector calculation step and the gain calculated by the gain calculation step to the corresponding base points;
A numerical control method comprising: an acquiring step of acquiring a command for moving a controlled object, which is at least one of a machine tool table and a tool;
a first filtering step of applying a first filter to a command trajectory, which is a movement trajectory when the controlled object moves according to the command acquired by the acquiring step;
a determining step of determining, as a point corresponding to the base point on the commanded trajectory, a corresponding point on the filtered trajectory that is the result of applying the first filter to the commanded trajectory in the first filtering step;
a vector calculation step of calculating, for each of the base points, an error vector indicating the distance and direction from each of the base points to each of the corresponding corresponding points;
a gain calculation step of calculating, for each base point, a gain corresponding to the error vector calculated by the vector calculation step;
a trajectory correction step of correcting the command trajectory by applying the error vector calculated by the vector calculation step and the gain calculated by the gain calculation step to the corresponding base points;
A storage medium storing a numerical control program for causing a computer to execute

Images