Control input generation device, control device, control input generation method, actuator, arithmetic circuit, control board, control input generation program, and storage medium storing control input generation program

Abstract

Provided is a control input generation device for generating a control input for controlling a pneumatic actuator with high precision. A device for generating a control input for controlling a pneumatic actuator is provided with: a first control input generation function for generating a first control input when there is a deviation of a control amount from a target value; a second control input generation function that generates, when there is a temporal change in the control amount, a second control input in which the deviation of the control amount from the target value becomes larger as the deviation becomes smaller; and an output generation function that generates the control input based on the first control input and the second control input. When the generated control input is used for controlling the pneumatic actuator, the responsiveness of the pneumatic actuator near the target value can be improved, and overshoot with respect to the target value can be suppressed, so that the accuracy of operating the pneumatic actuator to the target value can be improved. In addition, overshoot with respect to the target value can be prevented.

Description

TECHNICAL FIELD

[0001] The present application relates to a control input generation device, a control device, a control input generation method, an actuator, an arithmetic circuit, a control board, a control input generation program, and a control input generation program. More specifically, it relates to a control input generation device, a control device, a control input generation method, an actuator, an arithmetic circuit, a control board, a control input generation program, and a storage medium storing a control input generation program, which are suitable for the control of a pneumatic actuator using air pressure as a working fluid such as a cylinder. BACKGROUND

[0002] Various devices are employed for actuators that drive robots or mechatronic devices, and as actuators that utilize fluid pressure, hydraulic cylinders or air cylinders are generally used. For example, in a case where an object (an action object) that is moved by the actuator is moved to a destination position and disposed at the position, that is, in a case where the action object is moved by a prescribed amount, the actuator is controlled based on a difference (a deviation) between a current position of the action object and the destination position. That is, the current position of the action object is measured by a sensor, and the actuator is controlled in such a way that the deviation between the measured position and the destination position is 0.

[0003] In the actuator, the air cylinder is capable of driving the action object at high speed, and has the advantage that the output / weight ratio is high. In addition, compared with the hydraulic cylinder, it also has the advantages that the system constituting the actuator is relatively simple to construct and can be constructed at low cost, is explosion-proof, is clean and environmentally friendly, is easy to maintain, and the like.

[0004] On the other hand, the air cylinder has a problem in that the nonlinearity of the pressure response portion is stronger because the working fluid used is air having compressibility. In addition, there is a problem in that it is easily affected by the frictional force of each portion of the motion control system (for example, the Coulomb friction or the maximum static friction of the piston rod and the cylinder tube, and the like) due to the low rigidity characteristic. For these reasons, in the air cylinder, in a case where the action object is stopped at the destination position, there is a problem in that the positioning accuracy is inferior to that of other actuators such as the hydraulic cylinder, and in a control method using the existing PID control, it is difficult to perform micron-level action control using the air cylinder.

[0005] In the actuator using the air cylinder having the above technical problem, as a technique for improving the positioning accuracy thereof, there are the techniques disclosed in Non-Patent Literatures 1 and 2.

[0006] In Non-Patent Literatures 1 and 2, a device that moves an object on a plane by a cylinder is disclosed, and a technology that controls the driving of the cylinder based on a signal (position signal) obtained by sensing the position of the object, thereby controlling the movement amount of the object, is disclosed. In either of the technologies of Non-Patent Literatures 1 and 2, the influence of the maximum static friction force acting between the object and the plane on the control is suppressed, and thus the positioning accuracy is improved.

[0007] First, in Non-Patent Literature 1, a compensation signal that compensates for a main signal based on the difference between the current position obtained from the position signal and the target position is added to the main signal, and thus a control signal is formed. Specifically, the influence of the maximum static friction force on the control is suppressed by adjusting the compensation signal based on the natural frequency of the feedback control system. Further, it is described in Non-Patent Literature 1 that, by using this technology, even if the object is moved by the cylinder, the movement position of the object can be adjusted with respect to the target position with an accuracy of about 20 nm.

[0008] Further, in Non-Patent Literature 2, a technology that controls the driving of the cylinder by a pulse signal in which the pulse width is adjusted is disclosed. Further, it is described in Non-Patent Literature 2 that, by using this technology, even if the object is moved by the cylinder, the movement position of the object with respect to the target position can be adjusted with an accuracy of about 50 nm.

[0009] Prior Art Documents Non-Patent Literature Non-Patent Literature 1: Kei-Ren PAI and Ming-Qiang SHTH, nanoaccuracy Position Control of a Pneumatic Cylinder Driven Table, JSME International Journal, Series C, Vol. 46, No. 3, 1062-1067, 2003 Non-Patent Literature 2: Kaiji Sato, Yusuke Sano, Practical and intuitive controller design method for precision positioning of a pneumatic cylinder actuator stage, Precision Engineering 38 703-710, 2014 SUMMARY PROBLEMS TO BE SOLVED BY THE INVENTION However, in the technology of Non-Patent Literature 1, since the compensation signal is adjusted based on the natural frequency of the feedback control system, if the feedback control system cannot be grasped, high-precision position adjustment cannot be performed. Generally, the natural frequency of the feedback control system including the actuator changes due to the mass of the object moved by the actuator or the like, or the degradation of the actuator or the like over the years. Therefore, while the technology of Non-Patent Literature 1 can be applied to control in a laboratory experimental device, it is difficult to apply the control of the actuator or the like to an actual robot or mechatronic device or the like. In addition, it is presumed that Non-Patent Literature 1 adopts an air bearing in order to achieve positioning control while reducing the influence of the maximum static friction or the like as much as possible, and in this regard, it is also difficult to apply and perform high-precision position adjustment in an actual robot or mechatronic device or the like.

[0010] Furthermore, in the technology of Non-Patent Literature 2 as well, if the pulse width that enables appropriate positioning of the device including the cylinder cannot be grasped, high-precision position adjustment cannot be performed. The pulse width also changes due to the mass of the object moved by the actuator or the like, or the degradation of the actuator or the like over the years. Therefore, while the technology of Non-Patent Literature 2 can be applied to control in a laboratory experimental device, it is difficult to apply the control of the actuator or the like to an actual robot or mechatronic device or the like.

[0011] The present application has been made in view of the above circumstances, and aims to provide a control input generation device that generates a control input for high-precision control of a pneumatic actuator, a calculation circuit, and a control board.

[0012] Furthermore, the aim is to provide a control device that has a control input generation device that generates a control input for high-precision control of a pneumatic actuator, and an actuator that has the control device.

[0013] Furthermore, the aim is to provide a control input generation method that controls the driving of a pneumatic actuator with high precision, a control input generation program that executes the control input generation method, and a storage medium that stores the control input generation program.

[0014] Solution to the technical problem <Control input generation device> The control input generation device of the first aspect is a device that generates a control input for controlling a pneumatic actuator, characterized by having: a first control input generation function that generates a first control input in a case where there is a deviation of a control amount from a target value; a second control input generation function that generates a second control input that becomes larger as the deviation of the control amount from the target value becomes smaller in a case where there is a temporal change in the control amount; and an output generation function that generates the control input based on the first control input and the second control input.

[0015] The control input generating device according to a second aspect, in the first aspect, wherein the first control input generating function has a function of generating a constant steady-state control input as the first control input, or has a function of generating a control input in which the smaller the deviation of the control amount from the target value is, the smaller the fluctuation control input becomes, in combination with a constant steady-state control input, as the first control input, in the presence of a time variation of the control amount.

[0016] The control input generating device according to a third aspect, in the first aspect, wherein the second control input generating function calculates a predicted number of steps N until the control amount coincides with the target value, based on the deviation of the control amount from the target value and the operation state of the pneumatic actuator, and generates the second control input in a manner in which the smaller the predicted number of steps N is, the larger the second control input becomes.

[0017] The control input generating device according to a fourth aspect, in the second or third aspect, wherein the first control input generating function generates the first control input S1 based on Equation 1, the second control input generating function generates the second control input S2 based on Equation 2, and the output generating function generates the control input based on Equation 3.

[0018] Equation 1 S1 = K1 · sgn(e) Equation 2 S2 = K2 / N α · sgn(v) Equation 3 Yc = K1 · sgn(e) - K2 / N α · sgn(v) e: deviation of the control amount from the target value v: displacement speed of the pneumatic actuator a: power index for adjusting timing at which the suppression force that suppresses the operation of the pneumatic actuator starts to act K1: gain of the propulsion force K2: gain of the suppression force N: predicted number of steps The control input generating device according to a fifth aspect, in the fourth aspect, wherein the first control input generating function sets a gain K1 of the propulsion force that compensates for the nonlinear friction of the pneumatic actuator based on Equation 2-1, and the second control input generating function sets a gain K2 of the suppression force that reduces overshoot with respect to the target value based on Equation 2-2.

[0019] Equation 2-1 K1 = Ka |e| > Eth K1=Ka / (Eth·|e|) if |e|≤Eth Ka: prescribed control input Eth: prescribed displacement |e|: absolute value of the deviation e Equation 2-2 K2=Kb if |v|>Vth K2=0 if |v|≤Vth Kb: prescribed control input Vth: prescribed displacement velocity |v|: absolute value of the velocity v The control input generation device according to a sixth aspect, in the third aspect, wherein the second control input generation function has: a time-to-arrival calculation function that calculates a time Ta required for the control amount to coincide with the target value based on a deviation of the control amount from the target value; and a predicted step number calculation function that calculates a predicted step number N = Ta / △t obtained by dividing the time Ta by a sampling period △t at which the deviation of the control amount from the target value is detected, the predicted step number calculation function setting the predicted step number N to a threshold value if the predicted step number N is less than the threshold value.

[0020] The control input generation device according to a seventh aspect, in the sixth aspect, wherein the time-to-arrival calculation function calculates the time-to-arrival Ta based on Equation 4.

[0021] Equation 4 (1 / 2) · a · Ta 2 + v · Ta - e = 0 a: displacement acceleration of the pneumatic actuator v: displacement velocity of the pneumatic actuator Ta: time-to-arrival (time at which arrival - current time) e: deviation (target position - current position) The control input generation device according to an eighth aspect, in the third aspect, wherein the second control input generation function calculates a predicted step number N based on Equation 5 below.

[0022] Equation 5 N = e / △L = e / (v ·△t) e: deviation of the control amount from the target value v: displacement velocity of the pneumatic actuator △L: amount of change in displacement from 1 sampling time ago The control input generation device according to a ninth aspect, in the first aspect, wherein the target value is a target position of an object that is to be moved by the pneumatic actuator, or a driving velocity of the pneumatic actuator, or a pressure generated by the pneumatic actuator.

[0023] <Control device> The control device of the 10th aspect is a control device that controls a pneumatic actuator, characterized by having the control input generating device of the 1st aspect.

[0024] The control device of the 11th aspect is a control device that controls a pneumatic actuator, characterized by having a main control input generating section that generates a main control input that controls the pneumatic actuator, and a compensation control input generating section that generates a compensation control input that corrects the main control input, the compensation control input generating section being the control input generating device of the 1st aspect.

[0025] The control device of the 12th aspect is characterized in that, in the 10th or 11th aspect, the control device is a control system of Type 2 that has a disturbance observer.

[0026] The control device of the 13th aspect is characterized in that, in the 12th aspect, the control device implements feedforward control.

[0027] <Actuator> The actuator of the 14th aspect is characterized by having an operation section that operates using air pressure, and a control section that controls the operation of the operation section, the control section being the control device of the 10th or 11th aspect.

[0028] <Operation circuit> The operation circuit of the 15th aspect is a circuit that inputs a target value and a control amount as input signals, characterized by having: a first control input generating section that generates a first control input in the presence of a deviation of the control amount from the target value; a second control input generating section that generates a second control input that becomes larger the smaller the deviation of the control amount from the target value in the presence of a temporal change in the control amount; and an output generating section that generates and outputs a control input based on the first control input and the second control input.

[0029] <Control board> The control board of the 16th aspect is a control board having input terminals that input a target value and a control amount as input signals, and output terminals that generate and output a control input, characterized in that, in the control board, a main controller that generates a main control input based on signals input from the input terminals, a compensator that generates a compensation control input based on signals input from the input terminals, and an output generation section that generates a control input by combining the main control input and the compensation control input and supplies it to the output terminals, are provided, the compensator has a first control input generation function that generates a first control input in the presence of a deviation of the control amount from the target value, a second control input generation function that generates a second control input that becomes larger the smaller the deviation of the control amount from the target value in the presence of a temporal change in the control amount, and an output generation function that generates and outputs a control input based on the first control input and the second control input.

[0030] The control board of the 17th aspect is a control board having input terminals that input a target value and a control amount as input signals, and output terminals that generate and output a control input, characterized in that, in the control board, a compensator that generates a compensation control input based on signals input from the input terminals, and an output generation section that generates a control input by combining the compensation control input and a control input supplied from the outside and supplies it to the output terminals, are provided, the compensator has a first control input generation function that generates a first control input in the presence of a deviation of the control amount from the target value, a second control input generation function that generates a second control input that becomes larger the smaller the deviation of the control amount from the target value in the presence of a temporal change in the control amount, and an output generation function that generates and outputs a control input based on the first control input and the second control input.

[0031] <Control input generation method> The control input generation method of the 18th aspect is a method of generating a control input for controlling a pneumatic actuator, characterized in that a first control input is generated in the presence of a deviation of a control amount from a target value, a second control input that becomes larger the smaller the deviation of the control amount from the target value is generated in the presence of a temporal change in the control amount, and the control input is generated based on the first control input and the second control input.

[0032] The control input generation method of the 19th aspect is characterized in that, in the 18th aspect, the first control input is a constant steady-state control input, or, in the presence of a temporal change in the control amount, the first control input is a control input that combines a variable control input that becomes smaller the smaller the deviation of the control amount from the target value and a constant steady-state control input.

[0033] The control input generation method according to the 20th aspect, wherein in the 18th aspect, a predicted number of steps N until the control amount coincides with the target value is calculated based on a deviation of the control amount from the target value and an operation state of the pneumatic actuator, and the second control input is generated based on the predicted number of steps N in such a manner that the smaller the predicted number of steps N is, the larger the second control input becomes.

[0034] The control input generation method according to the 21st aspect, wherein in the 18th or 19th aspect, the first control input S1 is generated based on Equation 1, the second control input S2 is generated based on Equation 2, and the control input is generated based on Equation 3.

[0035] Equation 1 S1 = K1 · sgn(e) Equation 2 S2 = K2 / N α · sgn(v) Equation 3 Yc = K1 · sgn(e) - K2 / N α · sgn(v) e: a deviation of a control amount from a target value v: a displacement speed of a pneumatic actuator a: a power index for adjusting a timing at which a suppression force that suppresses an operation of the pneumatic actuator starts to act K1: a gain of a propulsive force K2: a gain of a suppression force N: a predicted number of steps The control input generation method according to the 22nd aspect, wherein in the 21st aspect, a gain K1 of a propulsive force that compensates for a nonlinear friction of the pneumatic actuator is set based on Equation 2-1, and a gain K2 of a suppression force that reduces an overshoot with respect to the target value is set based on Equation 2-2.

[0036] Equation 2-1 K1 = Ka |e| > Eth K1 = Ka / (Eth · |e|) |e| ≤ Eth Ka: a prescribed control input Eth: a prescribed displacement |e|: an absolute value of the deviation e Equation 2-2 K2 = Kb |v| > Vth K2 = 0 |v| ≤ Vth Kb: a prescribed control input Vth: a prescribed displacement speed |v|: an absolute value of the speed v The control input generation method according to the aspect 23, wherein in the aspect 20, the time Ta required until the control amount coincides with the target value is calculated based on a deviation of the control amount from the target value, and a predicted number of steps N = Ta / △t is calculated, where Ta is divided by a sampling period △t at which the deviation of the control amount from the target value is detected, and in a case where the predicted number of steps N is smaller than a prescribed threshold value, the predicted number of steps N is set to the threshold value.

[0037] The control input generation method according to the aspect 24, wherein in the aspect 23, the time Ta to reach is calculated based on Equation 4.

[0038] Equation 4 (1 / 2) · a · Ta 2 + v · Ta - e = 0 a: displacement acceleration of the pneumatic actuator v: displacement velocity of the pneumatic actuator Ta: time to reach (time to reach - current time) e: deviation (target position - current position) The control input generation method according to the aspect 25, wherein in the aspect 20, the second control input generation function calculates a predicted number of steps N based on Equation 5 below.

[0039] Equation 5 N = e / △L = e / (v ·△t) e: deviation of the control amount from the target value v: displacement velocity of the pneumatic actuator △L: amount of change in displacement from 1 sampling time ago The control input generation method according to the aspect 26, wherein in the aspect 18, the target value is a target position of an object that is to be moved by the pneumatic actuator, or a drive speed of the pneumatic actuator, or a pressure generated by the pneumatic actuator.

[0040] The control input generation program according to the aspect 27, wherein the program causes a control device that controls an operation of a pneumatic actuator to execute the control input generation method according to any one of the aspects 18 to 25.

[0041] The storage medium according to the aspect 28, wherein the storage medium stores the control input generation program according to the aspect 27.

[0042] Effects of the Invention <Control Input Generation Device> According to the first to ninth aspects, if the generated control input is used for control of the pneumatic actuator, it is possible to improve the responsiveness of the pneumatic actuator around the target value, and to suppress overshoot with respect to the target value, and thus it is possible to improve the accuracy of causing the pneumatic actuator to operate to become the target value. Furthermore, it is possible to prevent overshoot with respect to the target value.

[0043] <Control device> According to the tenth and eleventh aspects, if used as a control device of a pneumatic actuator, it is possible to improve the responsiveness of the pneumatic actuator around the target value, and to suppress overshoot with respect to the target value, and thus it is possible to improve the accuracy of causing the pneumatic actuator to operate to become the target value. Furthermore, it is possible to prevent overshoot with respect to the target value.

[0044] According to the twelfth aspect, it is possible to further improve the accuracy of causing the pneumatic actuator to operate to become the target value even in the case where there is a load variation.

[0045] According to the thirteenth aspect, it is possible to cause the pneumatic actuator to move in a manner that coincides with the target trajectory.

[0046] <Actuator> According to the fourteenth aspect, since the responsiveness around the target value is high, and overshoot with respect to the target value is suppressed, it is possible to operate with high accuracy in a manner that becomes the target value.

[0047] <Operation circuit> According to the fifteenth aspect, by providing in an existing control device of a pneumatic actuator, it is possible to improve the responsiveness of the pneumatic actuator around the target value, and to suppress overshoot with respect to the target value, and thus it is possible to improve the accuracy of causing the pneumatic actuator to operate to become the target value.

[0048] <Control board> According to the sixteenth and seventeenth aspects, by providing in an existing control device of a pneumatic actuator, it is possible to improve the responsiveness of the pneumatic actuator around the target value, and to suppress overshoot with respect to the target value, and thus it is possible to improve the accuracy of causing the pneumatic actuator to operate to become the target value.

[0049] <Control input generation device> According to the eighteenth to twenty-sixth aspects, if the generated control input is used for control of the pneumatic actuator, it is possible to improve the responsiveness of the pneumatic actuator around the target value, and to suppress overshoot with respect to the target value, and thus it is possible to improve the accuracy of causing the pneumatic actuator to operate to become the target value. Furthermore, it is possible to prevent overshoot with respect to the target value.

[0050] <Control input generation program, storage medium storing the control input generation program> According to the 27th and 28th aspects, if a program is installed in the control device that controls the pneumatic actuator and the program is executed, the responsiveness of the pneumatic actuator around the target value can be improved, and overshoot with respect to the target value can be suppressed, so the precision with which the pneumatic actuator is caused to operate to the target value can be improved. Also, overshoot with respect to the target value can be prevented. BRIEF DESCRIPTION OF DRAWINGS

[0051] [ Figure 1 ] is a block diagram of the control system 1 having the control device 10 of the present embodiment.

[0052] [ Figure 2 ] is an action explanatory diagram of the cylinder 2 controlled by the control device 10 of the present embodiment.

[0053] [ Figure 3 ] is a diagram showing an example of a position control system having the control device 10 of the present embodiment.

[0054] [ Figure 4 ] is a table showing the relationship of the compensation input Yc with the deviation e and the velocity v.

[0055] [ Figure 5 ] is an explanatory diagram of the state change of the compensation input Yc until the object M is stationary at the position of the target value Lr.

[0056] [ Figure 6 ] is a flow explanatory diagram until the object M is stationary at the position of the target value Lr.

[0057] [ Figure 7 ] is a table showing the relationship of the control amount with the parameters.

[0058] [ Figure 8 ] is a diagram showing an example of the operation circuit of the present embodiment.

[0059] [ Figure 9 ] is a schematic explanatory diagram of a conventional positioning control system to which a disturbance observer is attached.

[0060] [ Figure 10 ] (A) is a photograph of an experimental device, and (B) is a circuit diagram of the experimental device.

[0061] [ Figure 11 ] is a diagram showing the experimental results of Comparative Example 1, (A) is a graph of the response between 3 and 6 seconds, (B) is a graph of the response in the initial step (between 3 and 6 seconds) of the experimental results of (A) and is a graph in which the scale of the position deviation is enlarged.

[0062] [ Figure 12is a graph showing the experimental results of Example 1, (A) is a graph of the response in the 3-6 seconds, (B) is a graph of the response in the initial step (3-6 seconds) of the experimental results of (A) and is a graph in which the scale showing the position deviation is enlarged.

[0063] [ Figure 13 ] is a graph showing the experimental results of Example 1, enlarged Figure 12 the range of 10.0s-10.2s of (A), showing the change in the force generated with respect to the deviation from the target, (B) shows the change in the control input with respect to the deviation from the target in the same time period as (A).

[0064] [ Figure 14 ] is a graph showing the experimental results of Example 2.

[0065] [ Figure 15 ] (A) is a photograph of the experimental apparatus of Example 3, (B) is a graph showing the experimental results of Example 3.

[0066] [ Figure 16 ] is a schematic explanatory view of the control panel B of the present embodiment.

[0067] [ Figure 17 ] is a photograph of the experimental apparatus of Example 4.

[0068] [ Figure 18 ] (A) is a schematic view of the experimental apparatus of Example 4, (B) is a graph showing the state of the load variation.

[0069] [ Figure 19 ] is a graph showing the experimental results of the experiment of (1) of Example 4.

[0070] [ Figure 20 ] is a graph showing the experimental results of the experiment of (2) of Example 4.

[0071] [ Figure 21 ] is a graph showing the experimental results of the experiment of (3) of Example 4.

[0072] [ Figure 22 ] is a graph showing an example of the experiment of setting a constant.

[0073] [ Figure 23 ] is a graph showing the experimental results in the case where the sensor detecting the position of the work table was changed in Example 1, (A) is a graph of the response in the 0-24 seconds, (B) is a graph of the response in the initial step (0-4 seconds) of the experimental results of (A) and is a graph in which the scale showing the position deviation is enlarged.

[0074] [ Figure 24Fig. 1 is a graph showing experimental results in the case where the sensor detecting the position of the worktable was changed in Example 1, (A) is a graph of the response during 0 to 24 seconds, and (B) is a graph of the response in the initial step (0 to 4 seconds) of the experimental results of (A) and is a graph in which the scale showing the positional deviation is enlarged.

[0075] [ Figure 25 Fig. 2 is a graph showing experimental results in the case where the sensor detecting the position of the worktable was changed in Example 2.

[0076] [ Figure 26 (A) is a graph showing the relationship between the gain K1 and the deviation e, and (B) is a graph showing the relationship between the gain K2 and the displacement velocity v. DETAILED DESCRIPTION

[0077] Hereinafter, an embodiment of the present application will be described based on the drawings.

[0078] The control input generation device of the present embodiment is a device that generates a control input for control of a pneumatic actuator, and is characterized by being capable of generating a control input that improves control accuracy.

[0079] Further, the control input generation method of the present embodiment is a method that generates a control input for control of a pneumatic actuator, and is characterized by being capable of generating a control input that improves control accuracy.

[0080] <ACTUATOR> The actuator (control target) controlled by the control input generation device of the present embodiment is not particularly limited as long as it operates by air pressure. For example, a pneumatic actuator such as a cylinder or a rotary actuator that operates by air pressure can be cited as the actuator controlled by the control input generation device of the present embodiment. In addition, the displacement velocity in the present specification means the moving speed of the piston rod in the cylinder and the angular velocity of the rotary shaft in the rotary actuator, and the displacement acceleration means the acceleration of the piston rod in the cylinder and the angular acceleration of the rotary shaft in the rotary actuator.

[0081] <DEVICE EMPLOYING ACTUATOR> The machine or device employing the control input generation device of the present embodiment is also not particularly limited. That is, the machine or device having the actuator controlled by the control input generation device of the present embodiment is also not particularly limited. For example, a robot arm, a single-axis conveyance device, a multi-axis XY worktable, or the like that operates by the above-described actuator corresponds to the machine or device employing the control input generation device of the present embodiment.

[0082] <TARGET VALUE> When the actuator is controlled by the control input generation device of this embodiment, a target value is set to cause the actuator to perform a predetermined action, but this target value is not particularly limited. For example, if the actuator moves an object, the target position when the object moves or remains stationary, or the object's speed when it moves, can be used as the target value. Furthermore, when the object moves or remains stationary, the force applied to the object by the actuator can also be used as the target value.

[0083] In the following description, we will use the case where an object is moved by a cylinder as an example, and the target value is the position of the object.

[0084] Furthermore, the following description explains the case where the actuator is controlled by applying the control input generated by the control input generation device of this embodiment as a compensation input to a general control device for controlling the actuator. That is, it explains the case where the actuator is controlled by a control device having a general control device as the main control input generation unit and the control input generation device of this embodiment as a compensation control input generation unit. This control device is equivalent to the control device of claim 11.

[0085] <Control device 10 of this embodiment> The control device of this embodiment will now be described.

[0086] Figure 1 Reference numeral 2 indicates the control system 1, which controls the cylinder to move the object by means of the control device 10 of this embodiment. Furthermore, in the control system 1, the object M is fixed to the rod 2r of the cylinder 2, and when the rod 2r moves (i.e., when the cylinder 2 extends or retracts), the object M moves together with the rod 2r.

[0087] like Figure 1 As shown, the cylinder 2a of cylinder 2 is airtightly separated into a first chamber h1 and a second chamber h2 by piston 2p. The first chamber h1 and the second chamber h2 in cylinder 2a are connected to the air pressure source 4 via pipe 3 and servo valve 5, respectively. The servo valve 5 is electrically connected to the control device 10 of this embodiment (hereinafter referred to as control device 10).

[0088] The control device 10 controls the movement of the servo valve 5, thereby controlling the movement of the cylinder 2. It has the function of inputting target values, i.e., information regarding the position or amount of movement of the object M (or the position where the object M remains stationary), from an input mechanism or the like. The control device 10 is supplied with signals from pressure sensors p1 and p2 that measure the pressure in the first chamber h1 and the second chamber h2 of the cylinder 2. Therefore, the control device 10 can detect the force generated by the cylinder 2 and its direction based on the differential pressure between the pressure sensors p1 and p2. Furthermore, the control device 10 is supplied with signals from a position sensor d that measures the position of the object M, i.e., the distance of the object M from a predetermined position. Therefore, the control device 10 can detect the current position of the object M, which is moved by the cylinder 2, based on the distance detected by the position sensor d.

[0089] Furthermore, the control device 10 has the following function: based on the target value input from the input mechanism, etc., and the current position of the object M or the force and direction generated by the cylinder 2 as determined by signals from various sensors, it generates a signal (control input) to control the operation of the cylinder 2, i.e., a control input supplied to the servo valve 5. Specifically, the control device 10 includes: a main control input generation unit 11, which is a general control device for controlling the operation of the servo valve 5; and a compensation control input generation unit 15. The main control input generation unit 11 has the function of generating a main input as a control input based on the input signal using known control methods, such as linear control theory (PID control) or nonlinear control theory (fuzzy control, neural networks, etc.). On the other hand, the compensation control input generation unit 15 has the function of generating a compensation input Yc to correct the main input generated by the main control input generation unit 11, as detailed later.

[0090] As described above, if a target value is input to the control device 10 from the input mechanism, the cylinder 2 is controlled as follows.

[0091] For example, such as Figure 2 As shown in (A), when no external force is applied to the object M, and there is a difference (deviation e) between the current position (control quantity L) of the object M and the target value Lr, the control device 10 identifies the deviation e based on the signal from the position sensor d and supplies a signal to the servo valve 5 to actuate the cylinder 2, causing the object M to move towards the target value Lr. Then, the servo valve 5 adjusts the flow rate of the pressurized air supplied to the cylinder 2, causing the cylinder 2 to extend.

[0092] like Figure 2As shown in (B), if the object M approaches the target value Lr through the extension of cylinder 2, the deviation e becomes smaller. Therefore, the control device 10 supplies a signal to the servo valve 5 to reduce the force generated by cylinder 2, i.e., the force that moves object M toward the target value Lr. Then, the servo valve 5 adjusts the flow rate of pressurized air supplied to cylinder 2, reducing the force generated by cylinder 2 and decreasing the extension speed of cylinder 2.

[0093] like Figure 2 As shown in (C), if object M reaches the target value Lr (deviation e becomes 0), the control device 10 supplies a signal to the servo valve 5 to stop the operation of cylinder 2. Then, the servo valve 5 stops supplying pressurized air to cylinder 2, the operation of cylinder 2 stops, and thus object M can be kept stationary at the target value Lr.

[0094] On the other hand, such as Figure 2 As shown in (D), if an external force FR is applied to object M, and the cylinder 2 stops operating when object M reaches the target value Lr, there is a possibility that object M may move away from the target value Lr due to the external force FR. Even in this case, if a deviation e is detected, the control device 10 supplies a signal to the servo valve 5 to make the cylinder 2 operate in a way that reduces the deviation e, thus enabling object M to remain stationary at the target value Lr.

[0095] <Compensation Control Input Generation Unit 15> In the case of cylinder 2, the compressibility of air or the friction between object M and other objects has a significant impact on control accuracy (i.e., the difference between the stopping position of object M and the target value Lr). Therefore, to improve control accuracy, the main control input generation unit 11 is generally configured as a control system using a disturbance observer. However, even if a control system using a disturbance observer is configured, it is difficult to improve control accuracy to a certain level when the main control input generation unit 11 generates the control input using a general control method. The reasons are as follows.

[0096] In such Figure 9 In the positioning control system shown, it is assumed that the controller (a device having the same function as the main control input generation unit 11) controls the cylinder 2 using a general control method (e.g., PD control). Figure 9 The positioning control system shown includes a force-generating control system that compensates for parameter variations caused by load changes or the nonlinearity of the pressure response and friction by applying disturbance observers to each component. Furthermore, Figure 9 In the disturbance observer, the controlled object P will also generate force. f and the control object P of the motion kThe effects of changes relative to the nominal model are also inferred as disturbances. Therefore, to ensure the stability of the control system, the disturbance is considered as the inferred disturbance D that passes through the low-pass filter Q. fe D pe It is input into the target force Fr or the control input U.

[0097] in addition, Figure 9 The parameters in the table are represented as follows.

[0098] F r : Target generation force U: Control Input P f The object of control that generates force P k The object controlled by motion P fn -1 Inverse nominal model of the controlled object that generates force P kn -1 Inverse nominal model of the controlled object of motion Q f : Targeting D p Low-pass filter with predicted value Q k : Targeting D f Low-pass filter with predicted value D f Friction or external force D p Disturbance to the pressure response unit D fe :D f Estimated value D pe :D p Estimated value If the cylinder 2 is controlled by such a positioning control system, even if the object M moves to the target value Lr, it will not stop at the target value Lr and will not overshoot, resulting in a typical stick-slip phenomenon. This phenomenon occurs because: although the generated force F can follow the target generated force Fr, even if there is a deviation (if... Figure 2 If (B) decreases, the deviation e between the target value Lr and the current position of object M remains unchanged, and the force Fr generated by the target does not change. This is believed to be because if the deviation e becomes extremely small, the braking effect caused by the differential action of the PD control weakens. Since the control system is type I and has integral characteristics, its response to changes in deviation e becomes sluggish. As a countermeasure to reduce this overshoot, one approach is to increase the differential gain K. d This method can handle extremely small variations in deviation 'e', ​​but in this case, it leads to a significant reduction in the stability of the transition state.

[0099] On the other hand, in the control device 10 of this embodiment, such as Figure 1 and Figure 3 As shown, a positioning control system equipped with a compensation control input generation unit 15 is configured. This compensation control input generation unit 15 generates a compensation input Yc that corrects the main input Ms generated by the main control input generation unit 11. That is, in the control device 10 of this embodiment, the compensation input Yc generated by the compensation control input generation unit 15 is added to the main input Ms (in Figure 3 The control input is the main input Ms plus the predicted perturbation Fr, which is used as the control input. By inputting the control input that includes the compensation input Yc, the stick-slip phenomenon of cylinder 2 is suppressed, and the positioning control accuracy of object M is improved. For example, in the case where only the control input (main input Ms or input with predicted perturbation added to main input Ms) generated by the main control input generation unit 11 of the control device 10 of this embodiment is used, that is, by inputting the control input that includes the compensation input Yc, the stick-slip phenomenon of cylinder 2 is suppressed, and the positioning control accuracy of object M is improved. Figure 9 When the control system operates, the positioning control accuracy for placing object M at the target position is at most about 10 μm. However, in the control device 10 of this embodiment, by adding the compensation input Yc generated by the compensation control input generation unit 15 to the main input Ms (or adding the input after speculative perturbation to the main input Ms), the positioning control accuracy for placing object M at the target position can be improved to an accuracy equal to or higher than the detection accuracy of the position sensor that detects the position of object M. For example, if the detection accuracy of the position sensor that detects the position of object M is about 50 nm, a positioning accuracy of about 50 nm can be obtained; if the detection accuracy of the position sensor is about 1 nm, a positioning accuracy of about 1 nm can be obtained.

[0100] The compensation control input generation unit 15 generates the compensation input Yc. Figure 1 As shown, it has a first control input generation function 16, a second control input generation function 17 and an output generation function 18, and has the function of generating a compensation input Yc by inputting a target value Lr, a control quantity L and a piston speed v.

[0101] Specifically, the system has the function of generating a compensation input Yc (Yc = S1 - S2) using a first control input S1 generated by a first control input generation function 16 and a second control input S2 generated by a second control input generation function 17. The first control input S1 is the input that promotes the movement of cylinder 2 towards the target value Lr (the so-called input that makes cylinder 2 perform the "throttle" function), and the second control input S2 is the input that inhibits the movement of cylinder 2 towards the target value Lr (the so-called input that makes cylinder 2 perform the "brake" function). By appropriately generating the first control input S1 and the second control input S2, the positioning control accuracy of placing object M at the target position can be improved to about the detection accuracy of the position sensor by using the compensation input Yc.

[0102] <First Control Input Generation Function 16> The first control input generation function 16 has the function of generating a first control input S1 when there is a deviation e between the target value Lr and the control quantity L. Specifically, the first control input generation function 16 has the function of generating a constant steady-state control input as the first control input S1. For example, the first control input generation function 16 generates the first control input S1 based on the following Equation 1.

[0103] Formula 1 S1 = K1·sgn(e) e: Deviation between target value Lr and control quantity L K1 = Gain that compensates for the nonlinear friction of cylinder 2 and generates propulsion force. In addition, sgn(e) is a function that outputs 1 when e > 0, -1 when e < 0, and 0 when e = 0.

[0104] <Second Control Input Generation Function 17> The second control input generation function 17 has the function of generating a second control input S2 that increases as the deviation e between the target value Lr and the control value L decreases, given a time variation in the control quantity L. Specifically, the second control input generation function 17 has a prediction step calculation function 17b that calculates the prediction step number N until the control quantity L and the target value Lr are consistent, based on the deviation e between the target value Lr and the control quantity L and the piston speed v. Furthermore, the second control input generation function 17 has the function of generating the second control input S2 based on the prediction step number N calculated by the prediction step calculation function 17b, in a manner that increases as the prediction step number N decreases.

[0105] For example, the second control input generation function 17 can generate the second control input S2 based on the following equation 2. Additionally, the displacement velocity v of cylinder 2 is related to... Figure 3 The pistons in the two pistons move at the same speed, v.

[0106] Formula 2 S2 = K2 / N α ·sgn(v) e: Deviation of the control quantity from the target value v: Operating speed of cylinder 2 α: The power exponent used to adjust the timing at which the suppressive force for inhibiting the movement of cylinder 2 begins to take effect. K2 = Gain that produces a reduction in overshoot relative to the target value Lr In addition, sgn(v) is a function that outputs 1 when v > 0, -1 when v < 0, and 0 when v = 0.

[0107] <Third Control Input Generation Function 18> The third control input generation function 18 has the function of generating a compensation input Yc based on the first control input S1 generated by the first control input generation function 16 and the second control input S2 generated by the second control input generation function 17. Furthermore, the third control input generation function 18 is equivalent to the output generation function described in claim 1 of the claims.

[0108] For example, the compensation control input generation unit 15 can generate the compensation input Yc based on the following formula 3.

[0109] Formula 3 Yc=K1·sgn(e)-K2 / N α ·sgn(v) e: Deviation between target value Lr and control quantity L v: Displacement speed of cylinder 2 α: The power exponent used to adjust the timing at which the suppressive force for inhibiting the movement of cylinder 2 begins to take effect. K1 = Gain that compensates for the nonlinear friction of cylinder 2 and generates propulsion force. K2 = Gain that produces a reduction in overshoot relative to the target value Lr As described above, since the first control input S1 generated by the first control input generation function 16 and the second control input S2 generated by the second control input generation function 17 are combined, and the compensation input Yc is generated by the third control input generation function 18, the following advantages are obtained.

[0110] First, when the first control input generation function 16 generates the first control input S1 using Equation 1 above, it becomes a switch control where the first control input S1 changes in three stages. Therefore, when a deviation e occurs between the target value Lr and the control quantity L, the compensation signal Yc, serving as the first control input S1, contains a steady-state value with the same sign as the deviation e; when the deviation e is 0, the first control input S1 is 0. Thus, by applying the compensation signal Yc, which contains a constant first control input S1 independent of the magnitude of the deviation e, to the main input Ms as the control input, the responsiveness in transient states can be improved. For example, since friction generated by the piston of cylinder 2 or object M can be compensated near the target value Lr, the responsiveness near the target value Lr can be improved.

[0111] Furthermore, when the second control input generation function 17 generates the second control input S2 using Equation 2 above, when speed is generated (i.e., when the piston of cylinder 2 is moving), the compensation signal Yc becomes the signal obtained by subtracting the second control input S2, which is inversely proportional to the prediction step number N, from the first control input S1. When the speed is 0, the second control input S2 is 0, therefore the first control input S1 becomes the compensation signal Yc. Thus, since the effect of suppressing the movement (applying brakes) of cylinder 2 near the target value Lr is enhanced, the occurrence of stick-slip can be suppressed, and a high-precision positioning effect at the target value Lr can be obtained.

[0112] As described above, in the control device 10 of this embodiment, since the compensation input Yc generated by the compensation control input generation unit 15 is added to the main input Ms generated by the main control input generation unit 11 as the control input, the responsiveness of the cylinder 2 in the transition state can be improved, and a high-precision positioning effect at the target value Lr can be obtained.

[0113] <State of the input Yc> Below, based on Figure 5 and Figure 6 This describes the process until object M comes to rest at the target value Lr (until control converges).

[0114] The compensation signal Yc is determined based on the deviation e between the target value Lr and the control quantity L, and the sign of the displacement velocity (piston velocity v) of cylinder 2. Figure 4 As shown in the table. Therefore, if the operation starts from the state where cylinder 2 is not activated (e.g., e > 0, v = 0), then the compensation signal Yc = K1, and cylinder 2 moves towards the target value Lr (positive direction). If cylinder 2 moves towards and approaches the target value Lr, the deviation e decreases, and soon, due to the second control input S2 (K2 / N... α Due to the influence of ), the speed of cylinder 2 decreases, and soon the action of cylinder 2 stops (v=0).

[0115] At this time, if the control quantity L of cylinder 2 is consistent with the target value Lr (e=0), then cylinder 2 will remain stationary. On the other hand, if there is a deviation e when cylinder 2 is stopped (e>0 when it stops just before reaching the target value Lr), cylinder 2 will start to move towards the target value Lr due to the influence of the first control input S1 (K1).

[0116] If the movement is directed toward the target value Lr, then due to the second control input S2 (K2 / N) α Due to the influence of ), the movement of cylinder 2 stops (v=0). At this time, when it stops just before reaching the target value Lr (e>0), the cylinder 2 continues to move in the positive direction and stop repeatedly until the control quantity L of cylinder 2 matches the target value Lr.

[0117] On the other hand, when the control quantity L of cylinder 2 exceeds the target value Lr (e < 0), the first control input S1 (K1) and the second control input S2 (K2 / N) α Both of these functions act as brakes, quickly stopping the movement of cylinder 2 (v = 0). In this case, due to the deviation e (e < 0), cylinder 2 begins to move towards the target value Lr due to the influence of the first control input S1 (K1). That is, it begins to move towards the target value Lr in the opposite direction to the initial movement direction (reverse movement).

[0118] If cylinder 2 starts to operate, then due to the second control input S2 (K2 / N) α Due to the influence of [various factors], the speed of cylinder 2 decreases, and soon the movement of cylinder 2 stops (v = 0). At this time, if the control quantity L of cylinder 2 is consistent with the target value Lr (e = 0), then cylinder 2 remains stationary. On the other hand, if it stops just before reaching the target value Lr (e < 0), the process of moving and stopping in the opposite direction is repeated until the control quantity L of cylinder 2 is consistent with the target value Lr.

[0119] Furthermore, during reverse movement, there are cases where the control quantity L of cylinder 2 becomes smaller than the target value Lr (e > 0). In this case, similar to the overshoot case, the first control input S1 (K1) and the second control input S2 (K2 / N) also become smaller. α Both of these functions act as brakes, quickly stopping the movement of cylinder 2 (v=0). In this case, due to the deviation e (e>0), cylinder 2 starts to move due to the influence of the first control input S1 (K1), repeatedly moving and stopping in the positive direction.

[0120] As described above, by using the compensation input Yc, cylinder 2 repeatedly moves and stops towards the target value Lr, thereby ensuring that the control quantity L matches the target value Lr. Furthermore, since the first control input S1 (K1) is constant when moving again after stopping, even with a small deviation e, cylinder 2 can move again at high speed from its stopped state. Additionally, because the second control input S2 (K2 / N) is near the target value Lr... α The value increases, thus suppressing stick-slip phenomenon and enabling high-precision positioning of the target value Lr.

[0121] <Regarding the first control input generation function 16> This explains the case where the first control input generation function 16 generates a constant steady-state control input as the first control input S1. However, the first control input generation function 16 can also generate a variable control input as the first control input S1 by adding a variable control input to the constant steady-state control input, where the deviation e of the control quantity L from the target value Lr becomes smaller as the deviation e becomes smaller. For example, in the initial stage of a transient response with a relatively large deviation, by setting it as the first control input S1 with the variable control input, advantages such as improved responsiveness can be obtained.

[0122] <Regarding the gain K1 of the first control input S1 and the gain K2 of the second control input S2> The gains K1 of the first control input S1 and K2 of the second control input S2 can be set to the same value regardless of the direction of movement of cylinder 2, but they can also be set to different values ​​depending on the direction of movement of cylinder 2. When cylinder 2 is a single-rod cylinder, due to the influence of the rod, the pressure area of ​​the piston is different on both sides of the piston. Therefore, in the case of a single-rod cylinder, the piston's pushing action (moving towards the opposite side of the rod) generates force more easily than the piston's pulling action (moving towards the rod side). That is, pushing is more likely to move the piston than pulling. Therefore, when cylinder 2 is a single-rod cylinder, the gains K1 of the first control input S1 and K2 of the second control input S2 can also be changed according to the direction of piston movement, as follows.

[0123] K1 = K1a (e > 0) K1 = K1b (e < 0) K2 = K2a (v > 0) K2 = K2b (v < 0) Where K1a < K1b and K2a < K2b.

[0124] Furthermore, when a force is applied to cylinder 2 in a single direction, it is also affected by this force. Therefore, even if the same amount or speed of movement is achieved according to the piston's direction of movement, the required force, or in other words, the differential pressure generated between the first chamber h1 and the second chamber h2 of cylinder 2, will be different. Therefore, when an external force is applied to cylinder 2 in the direction of pushing, the gain K1 of the first control input S1 and the gain K2 of the second control input S2 can be set in the same manner as described above. On the other hand, when an external force is applied to cylinder 2 in the direction of pulling, the gain K1 of the first control input S1 and the gain K2 of the second control input S2 can be set in the order K1a > K1b and K2a > K2b.

[0125] Furthermore, the gain K1 of the first control input S1 and the gain K2 of the second control input S2 can be set to fixed values ​​(constants), but they can also be configured to vary according to the actual deviation or displacement velocity. Thus, if the gain K1 of the first control input S1 is set as in Equation 2-1, a specified input that promotes movement toward the target value can be generated until the deviation e reaches the required positioning accuracy (required deviation). On the other hand, if the absolute value |e| of the deviation e becomes less than the required deviation, it can be set as an input that attenuates the movement, thereby further improving the responsiveness near the target value.

[0126] Furthermore, if the gain K2 of the second control input S2 is set as in Equation 2-2, and the gain K2 is set to 0 when the displacement speed becomes less than the required displacement speed (required displacement speed), it can prevent the second control input S2 from becoming unstable in the extremely low speed domain where the influence of noise is greatly amplified.

[0127] Equation 2-1 The case where K1 = Ka|e| > Eth K1 = Ka / (Eth·|e|) (case where |e|≤Eth) Ka: The minimum control input to produce displacement Eth: Required positioning accuracy (required deviation) (deviation threshold) |e|: The absolute value of the deviation e Equation 2-2 K2=Kb|v|>Vth K2=0|v|≤Vth Kb: The minimum control input to generate displacement Vth: Threshold for displacement velocity taking into account the effect of observation noise. |v|: The absolute value of velocity v If the gain K1 of the first control input S1 is set as in Equation 2-1, then the gain K1, the first control input S1, and the deviation e become as follows: Figure 26The relationship is as shown in (A). Furthermore, if the gain K2 of the second control input S2 is set as in Equation 2-2, then the gain K2 and the second control input S2 and displacement velocity v become as shown in Equation 2-2. Figure 26 The relationship shown in (B).

[0128] In this invention, based on the following technical concept, the compensation input Yc is generated by the first control input S1 and the second control input S2 as described above.

[0129] First, using Equations 2-1 and 2-2 above, the gain K1 of the first control input S1 and the gain K2 of the second control input S2 are set. When the compensation input Yc is generated based on Equation 3, the gain K1 of the first control input S1 (the first term) is designed to exceed the gain of the maximum static friction force. Therefore, the first control input S1 is designed as a control input that, under any conditions and environment, causes the controlled object to move without stopping its displacement until the deviation reaches a predetermined threshold (until the absolute value of the deviation becomes less than the predetermined threshold). During the process of reaching the state where the absolute value of the deviation of the controlled object becomes the predetermined threshold, even when the displacement of the controlled object is about to stop, by making the second control input S2 (the second term) a value close to 0, the controlled object can continue to move in such a way that the absolute value of the deviation becomes the predetermined threshold. That is, when the absolute value of the deviation is close to the specified threshold and the controlled object is about to stop, because the prediction step number N increases (in other words, the change in displacement ΔL from one sampling moment ago becomes a smaller value), the second control input S2, which applies braking force to the movement of the controlled object in the second term of the compensation input Yc, becomes close to 0. Thus, the compensation input Yc essentially becomes a state with only the first term, so the controlled object can overcome friction and continue to move.

[0130] On the other hand, the second control input S2 is designed to: when the deviation is within a specified threshold (the absolute value of the deviation becomes less than the specified threshold) and the target value (e=0) is reached, if the speed (absolute value of the speed) at this time exceeds the specified threshold, then the controlled object will generate a braking force to stop the controlled object's action, thereby performing a stop displacement action.

[0131] Furthermore, if the first control input S1 and the second control input S2 are designed as described above, then when the deviation is within a predetermined threshold (when the absolute value of the deviation becomes less than the predetermined threshold), the function of the first control input S1 (as described in item 1) allows for an immediate return to within the threshold even when the deviation is outside the predetermined threshold (when the absolute value of the deviation becomes greater than the predetermined threshold). Since the main controller generally has an integrator, it outputs a signal that integrates the deviation to set it to 0 whenever the deviation is not 0. Thus, although it is possible for the deviation to deviate from the predetermined threshold, the effect of the first control input S1 (as described in item 1) allows for a rapid return of the deviation to within the threshold.

[0132] Furthermore, if the first control input S1 and the second control input S2 are designed as described above, even if an external force is applied and the speed (absolute value of the speed) of the controlled object exceeds a predetermined threshold, and the deviation is outside the predetermined threshold (the absolute value of the deviation becomes greater than the predetermined threshold), the function of the second control input S2 (as item 2) can apply a larger braking force to the controlled object to prevent the deviation from deviating significantly from the threshold. As a result, the effect of the first control input S1 (item 1) can quickly bring the deviation back within the threshold.

[0133] <Regarding the prediction step calculation function 17b> The method for calculating the prediction step number N in the prediction step calculation function 17b of the second control input generation function 17 is not particularly limited. For example, the prediction step number N can also be the time required to make the control quantity L consistent with the target value Lr by using the sampling period Δt of the deviation e between the detected target value Lr and the control quantity L (e.g., the period during which the position sensor detects the control quantity L). That is, as... Figure 1 As shown, an arrival time calculation function 17a can also be set in the second control input generation function 17 to calculate the time (arrival time Ta) from the current position to the target position. The arrival time Ta calculated by the arrival time calculation function 17a is divided by the sampling period Δt, and the value obtained (i.e., N = Ta / Δt) is used as the prediction step number N. If the prediction step number calculation function 17b calculates the prediction step number N by this method, the advantage of making the calculation of the prediction step number N easier is also obtained. In addition, N = Ta / Δt can also be calculated by the following equation 5. In addition, v·Δt is the same as the movement amount ΔL per sampling period (in other words, the change in displacement since one sampling moment).

[0134] Formula 5 N(e,v)=e / △L=e / (v·△t) e: Deviation between target value Lr and control quantity L v: Displacement speed of cylinder 2 △L: The change in displacement since one sampling time point Furthermore, N in Equation 5 can also be obtained through Equation 5-1 below.

[0135] Formula 5-1 N(Lr,L)=(Lr(k)-L(k)) / (L(k)-L(k-1)) Lr(k): Target position L(k): Current position L(k-1): Current position one step ago Furthermore, the prediction step calculation function 17b may also have the function of setting the prediction step N to the threshold X when the prediction step N is less than a predetermined threshold X. For example, if the prediction step N becomes less than 1 when the threshold X is set to 1, the prediction step calculation function 17b may also have the function of setting the prediction step N to 1.

[0136] The reason for setting up this function is as follows. For example, the second control input S2 has the gain (K2 / N) described above. α In the case of [condition], if the piston speed v is too large relative to the deviation e between the target value Lr and the control quantity L, the second control input S2 becomes extremely large, potentially diverging to infinity, leading to a loss of control over the control system. Therefore, to limit the value of the second control input S2 and prevent the control system from going out of control, it is desirable to set the minimum value of the prediction step number N as the threshold X.

[0137] Furthermore, while the example above illustrates the generation of the second control input S2 using Equation 2, the method by which the prediction step calculation function 17b generates the second control input S2, i.e., the formula for generating the second control input S2 (the second control input generator formula), is not limited to Equation 2. It is sufficient that the second control input S2 is generated in a manner that increases as the prediction step number N decreases. In other words, it is sufficient that the second control input S2 can be generated such that the actuator's action can be appropriately converged using the compensation input Yc generated based on the first control input S1 and the second control input S2. That is, a second control input generator formula is used to generate the second control input S2 that enables the convergence of the actuator's action to be in the desired state.

[0138] <Calculation method for arrival time Ta> In the second control input generation function 17, the method for calculating the arrival time Ta by the arrival time calculation function 17a, which calculates the time required for the control quantity L to reach the target value Lr, is not particularly limited. For example, in the following equation 4, a and v are taken as known values, and the arrival time Ta can be calculated using Newton's method, for example.

[0139] Formula 4 (1 / 2)·a·Ta 2 +v·Ta-e=0 a: Displacement acceleration of cylinder 2 v: Displacement speed of cylinder 2 Ta: Arrival Time (Arrival Time - Current Time) e: Deviation (target position - current position) <Regarding the third control input generation function 18> In the above example, the case where the third control input generation function 18 generates the compensation input Yc based on Equation 3 was illustrated. However, the formula (compensation input generation formula) for the third control input generation function 18 to generate the compensation input Yc is not particularly limited. A compensation input generation formula that can generate a compensation input Yc that can appropriately converge the actuator action based on the first control input S1 and the second control input S2 can be used. That is, a compensation input generation formula that can generate a compensation input Yc that makes the convergence of the actuator action reach the desired state based on the first control input S1 and the second control input S2 is acceptable.

[0140] <The case where the control input generation device of this embodiment is used as the control device> In the above example, the case where the control input generation device of this embodiment is included in the control device as a compensation input generation device was illustrated. However, the operation of the cylinder 2 can also be controlled solely by the control input generation device of this embodiment. That is, the control input generation device of this embodiment can be used as a control device to control the operation of the cylinder 2. In this case, the compensation input generated by the control input generation device of this embodiment is input as the control input to control the operation of the cylinder 2.

[0141] <Operational Circuits> The control input generation device of this embodiment can also be provided as an arithmetic circuit. That is, an arithmetic circuit can be manufactured, and the control input generation device of this embodiment can be provided as such (the arithmetic circuit of this embodiment). This arithmetic circuit has input terminals that input a target value and a current value as input signals, and output terminals that output a signal equivalent to the control input generated by the control input generation device of this embodiment as an output signal.

[0142] That is, a circuit capable of being formed and provided as the operational circuit of this embodiment is provided. This circuit includes a first control input generation unit, a second control input generation unit, and an output generation unit. When a target value and a control quantity are input signals, if there is a deviation between the control quantity and the target value, the first control input generation unit generates a first control input; if there is a time variation in the control quantity, the second control input generation unit generates a second control input that increases as the deviation between the control quantity and the target value decreases; the output generation unit generates and outputs a control input based on the first and second control inputs. For example, if it is set as follows... Figure 8 The circuit shown can be configured to form an arithmetic circuit of this embodiment with the above-described functions, and therefore a circuit with this configuration can be provided as an arithmetic circuit of this embodiment.

[0143] When the control input generation device of this embodiment is provided in the form of an arithmetic circuit, if an existing control device is used directly and the control input generated by the control input generation device of this embodiment is incorporated into the existing control device in such a way as a compensation input, the existing control device with the existing actuator can perform the same function as the control device of this embodiment.

[0144] Furthermore, as described above, the arithmetic circuit can either supply compensation input to an existing control device or be configured such that the control input of the existing control device is also input into the arithmetic circuit, and the arithmetic circuit combines the control input of the existing control device with the compensation input (the control input generated by the control input generation device in this embodiment) to output a control input.

[0145] <Control Panel> The control input generation device of this embodiment can also be provided as a control board. That is, a control board can be manufactured and provided as a control board, which has input terminals for inputting target values ​​and current values ​​as input signals, and output terminals for outputting signals equivalent to the control input generated by the control input generation device of this embodiment.

[0146] For example, in Figure 16 The control panel B shown in (A) has input target value and control quantity (in Figure 16The input terminal IT in (A) represents signals from a pressure sensor and signals from a position sensor. It also includes: a main controller MC, which generates a control input (main control input) equivalent to that of a general control device based on the input target value and control quantity; and a compensator CM, which has the same function as the aforementioned compensating control input generation unit 15 (i.e., it has the function of generating compensating control input). Furthermore, the control board B includes: an output generation unit D3, which is an addition circuit that combines the main control input and the compensating control input (addition operation) to generate a control input; and an output terminal OT, which outputs the control input generated by the output generation unit D3 to the outside. By providing such a control board B, by replacing the existing control device (control board) with the control board B, control based on the control input generation method of this embodiment can be introduced into the control of existing equipment.

[0147] In addition, Figure 16 In (A), although the main control input generated by the main controller MC is configured to be input from the input terminal and supplied to the output generation unit D3 after being output from the output terminal OT, the main control input generated by the main controller MC can also be configured to be input directly to the output generation unit D3 without being output from the output terminal OT. Furthermore, in Figure 16 In the case of configuration (A), the main control input output from the output terminal OT, that is, without being combined with the compensation control input, can be supplied to the controlled object as a control input, thereby increasing the control freedom of the controlled object.

[0148] Furthermore, when incorporating the control input generation method based on this embodiment into the control of existing equipment, control board B can also utilize the control input output by the existing control device EM, i.e., the control input from an external source, to generate control input including compensation input. For example, as in Figure 16 In the control board B shown in (B), in addition to the input terminal for inputting target values ​​or control quantities, an input terminal IT is also provided for inputting control inputs from the control device EM of an existing device. Furthermore, the output generation unit D3 can also use the control input from this input terminal IT as the main control input to generate control inputs. If such a control board B is provided, control based on the control input generation method of this embodiment can be introduced into the existing control device EM simply by connecting the control board B to the existing control device EM.

[0149] <Control Input Generation Program> Furthermore, in the aforementioned example, it was explained that by providing the control input generation device of this embodiment in the state of a control panel, control based on the control input generation method of this embodiment can be imported. However, control based on the control input generation method of this embodiment can also be imported by implementing a program that implements the control input generation method of this embodiment.

[0150] That is, by installing a program that implements the control input generation method of this embodiment into the actuator that becomes the controlled object or the control device (computer, etc.) that employs the actuator, or by providing the program via an internet connection, it is also possible to import control based on the control input generation method of this embodiment onto the controlled object. When installing a program into the control device, the program can be installed either by providing the program via an internet connection or by storing the program in a storage medium readable by the control device and installing the program from that storage medium.

[0151] For example, a program (the control input generation program of this embodiment) is installed on the control equipment of an existing machine or device to implement a first control input generation function, a second control input generation function, and an output generation function. When a target value and a control quantity are input signals, the first control input generation function generates a first control input if there is a deviation between the control quantity and the target value. The second control input generation function generates a second control input that becomes larger as the deviation between the control quantity and the target value decreases over time if there is a change in the control quantity. The output generation function generates and outputs a control input based on the first and second control inputs. Furthermore, if the compensation input generated by the control input generation program of this embodiment is attached to the control input (main control input) generated by the control equipment of the existing machine or device, control performed by the control input generation method of this embodiment can be implemented in the existing machine or device.

[0152] Furthermore, the control input generation program of this embodiment can also have the function of generating control inputs equivalent to the main control input. That is, the control input generation program of this embodiment can also have the function of generating the main control input, the function of generating compensation input, and the function of generating control inputs based on the main control input and the compensation input. In this case, if the control input generation program of this embodiment is installed in the control equipment of an existing machine or device instead of the existing control program, control based on the control input generation method of this embodiment can be introduced into the control equipment of the existing machine or device.

[0153] Storage media for storing the control input generation program of this embodiment can include, for example, floppy disks, hard disks, optical disks, magneto-optical disks, CD-ROMs (Compact Disk-ROMs), magnetic tapes, non-volatile memory cards, and ROMs. Furthermore, programs that are not limited to being executed independently by a program, but also those that run in conjunction with other software and expansion board functions on an OS (Operating System), are also included within the scope of each embodiment.

[0154] <Regarding control quantity> In the above example, the case where the control quantity controlled by the control input or compensation input generated by the control input generation device of this embodiment is the movement amount of cylinder 2 was described; that is, the case of a control system for position control performed by the control input generation device of this embodiment was described. The control quantity controlled by the control input or compensation input generated by the control input generation device of this embodiment is not limited to the position described above. For example, if we set it to the following Equation 6, so that each parameter corresponds to... Figure 7 The table shows that the displacement speed of cylinder 2 or the force generated by cylinder 2 can also be controlled by generating a compensation input Yc (or control input) in the same way as position control.

[0155] Formula 6 Yc=K1·sgn(e)-K2 / N α ·sgn(v) N = e / (v·△t) = (x r (k)-x(k)) / (x(k)-x(k-1)) xr(k): Target value x(k): Current value x(k-1): The current value one step ago. Furthermore, N in Equation 6 can also be obtained through Equation 6-1 below.

[0156] Formula 6-1 N(xr,x)=(xr(k)-x(k)) / (x(k)-x(k-1)) xr(k): Target value x(k): Current value x(k-1): The current value one step ago. For example, if used in Figure 3 The force control system internally configured within the position control system shown can use the control input generation device of this embodiment as a device for generating compensation input in the control system that controls the force generated by the cylinder 2 as described above.

[0157] <Regarding the conjecture of constants> As mentioned earlier, in generating the formula for the compensation input Yc in the compensation control input generation unit 15, various constants such as gain need to be set. These constants are usually determined through repeated experiments. That is, while changing the constants, the control object (a device that moves by means of a cylinder or the like) is operated using the control input generation device 1 of this embodiment to confirm the operation status, and the constants are changed in such a way as to achieve an appropriate operation status, and the final constants are set.

[0158] For example, if the compensation input is generated by Equation 3 above, the controlled object is made to act by the compensation input generated according to the set constants such as gain K1, K2, and power exponent α, and the result of the action is evaluated. The combination of constants that obtains an appropriate evaluation is finally set as a constant.

[0159] However, while it is possible to find an appropriate constant in a relatively short time when setting constants using this method, it usually requires a very large number of experiments to find the constant that can achieve the desired motion accuracy.

[0160] Therefore, if a predictive device capable of predicting constants using machine learning models is developed, then even with a small number of experiments, appropriate constants can be predicted based on the constants used in the experiment and the results of the actions.

[0161] For example, an evaluation function is set to assess the outcome of actions performed when a controlled object is made to move. Learning data is collected that correlates the constants set when the controlled object is made to move, the evaluation result (assessment result) after evaluating the controlled object's actions under the set constants using the evaluation function, and the evaluation of experimental results. Furthermore, if a machine learning model, which serves as a prediction device, learns from the learning data, an evaluation result can be obtained simply by inputting constants into the prediction device, even without conducting experiments using the controlled object.

[0162] There are no particular restrictions on the machine learning model used by the inference device; it can employ support vector machines (SVM), neural networks (NN), random forests (RF), etc.

[0163] The evaluation function for assessing motion accuracy described above can be adapted to the controlled object or its control quantity. For example, if the object is cylinder position control, the motion accuracy can be evaluated using the evaluation function according to Equation 7 as follows (see Equation 7). Figure 22 In this case, it means that the smaller the evaluation score S1, the more appropriate the position control is (the target trajectory R is consistent with the actual action A).

[0164] Formula 7 S1 = w1·OR + w2·PA S1: Evaluation Score OR: The ratio of overshoot (OS) to target movement (DA) (OS / DA) PA: Position accuracy polarity ACC > E: 1 ACC≤E: 0 w1, w2: Weights For position accuracy polarity PA, when the actual action A becomes a stable state, the difference ACC between the target movement DA and the actual movement RA in the stable state is set to 0 if it falls within the specified value E, and set to 1 if it is greater than it.

[0165] In addition, after implementing such Figure 22 In the case of the action command shown (target trajectory R), Equation 8 can also be used as the evaluation function. In this evaluation function, it also means that the smaller the evaluation score S2, the more appropriate the position control is (the target trajectory R is consistent with the actual action A).

[0166] Formula 8 S2 = w3·OS + w4·ACC + w5·TM S1: Evaluation Score OS: Overshoot relative to the target movement DA ACC: The difference between the target movement DA and the actual movement RA in the state that becomes a steady state. TM: Time from the start of control until the movement stabilizes w3, w4, w5: Weights <Regarding Control Systems as Objects> By employing the control input generation device 1 of this embodiment, the positioning control accuracy of placing the object M at the target position can be improved to the detection accuracy of a position sensor. On the other hand, if there are changes in the external environment affecting the movement of the object M, the positioning control accuracy may decrease in response to the changes in the external environment (transient response). However, if the control input generation device 1 of this embodiment is configured as a Type 2 control system with a disturbance observer, and feedforward control is implemented in a way that eliminates its dynamics, the tracking accuracy of the target trajectory during the transient response can be improved.

[0167] The method for designing the control input generation device 1 of this embodiment to construct a Type 2 control system with a disturbance observer is not particularly limited. For example, if it is Figure 3 The position control system shown above, if Gf(s) / (1-Q(s)·Gf(s)) (Gf(s) is the transfer function of the force control system) independently contains 1 / s 2By designing the control system in the manner of (1 / s is an integrator), the control input generation device 1 of this embodiment can be set as a type 2 control system with a disturbance observer.

[0168] Thus, when the control input generation device 1 of this embodiment is designed as a type 2 control system with a disturbance observer, even if changes in the external environment affecting the movement of object M occur, the decrease in tracking accuracy of the target trajectory due to the transient response caused by the change in the external environment can be suppressed. That is, when controlling the movement of object M, even if changes in the external environment, such as load changes, occur midway, stable movement of object M can be maintained.

[0169] Example The effect of applying the compensator (the compensating control input generation unit of the present invention), which generates compensating inputs using the control input generation method of the present invention, to a control device using a general control method has been confirmed.

[0170] The experiment is in the construction Figure 10 The air pressure drive circuit shown in (B) Figure 10 The experiment was conducted in the apparatus shown in (A) and the accuracy of positioning the worktable in the specified position by the action of the cylinder was confirmed.

[0171] The cylinder used in the experimental setup is a general-purpose double-acting cylinder (SMC, CDJ2E16-150Z-B, 16mm inner diameter, 150mm rod length). The worktable is supported by linear guides (THK, SHW12CA1UUM190LM) and is configured to move under the guidance of the linear guides.

[0172] The double-acting cylinder is driven by a flow-controlled servo valve (Festo, MPYE-5-M5-010-B). Specifically, the pressure in both cylinder chambers is measured by a pressure sensor (Nidec Components, PA-500-103G), and the pressure difference between the two chambers is adjusted by the servo valve based on this measurement. Furthermore, the air supply pressure to the servo valve is 0.3 MPa.

[0173] The position of the worktable is determined by measuring the displacement of the rod of the double-acting cylinder using a linear encoder (iC-Haus, iC-PZ205 EVAL PZ1M_Z, resolution 12.5nm).

[0174] The control system that generates the control signal supplied to the servo valve is a system that has, for example, the control signal supplied to the servo valve. Figure 3The type 1 feedforward control system of the disturbance observer shown is built on a PC using RTAI, a real-time extension of the Linux kernel. The acquisition of pressure data and the application of voltage to the servo valve by the control system are performed via SPI communication between various IC chips and the Raspberry Pi 4B. Furthermore, the pressure data is acquired by performing A / D conversion on the signal supplied from the pressure sensor, and the voltage application to the servo valve is performed by performing D / A conversion on the control signal generated by the control system. Additionally, the displacement data from the linear encoder is acquired via UART communication after being filtered to remove noise by an STM32 microcontroller.

[0175] In addition, since the sampling period of the control system is 2ms, the Raspberry Pi 4B communicates with the PC via UDP-based sockets, and the process from A / D conversion request to conversion value acquisition takes 0.5ms, thus enabling it to perform real-time tasks effectively.

[0176] Furthermore, because noise was mixed into the displacement data from the linear encoder used in this experiment, the deviation between the target value and the measured value was not strictly zero. Therefore, the compensator continuously outputs compensation input, causing chattering near the target value. As a countermeasure, a threshold for the deviation was set in the control system used in this experiment; if the deviation was below the threshold, no compensation input was output.

[0177] Furthermore, in the extremely low-speed range, the second control input generation unit of the compensator becomes unstable due to the increased influence of noise applied to the speed parameters. As a countermeasure, in the control system used in this experiment, a speed threshold is set; if the speed is below the threshold, the second control input is not output.

[0178] <Example 1> The control performance of the control input generation method of the present invention was verified when the compensation input generated by the compensator using the control input generation method of the present invention was incorporated into the target force and the control input. In the experiment, the response (position deviation) when repeatedly applying step inputs of 10 mm and 50 mm as target values ​​was analyzed, along with the case where the compensation input generated by the control input generation method of the present invention was incorporated (Example 1, Reference). Figure 3 ), and the case without compensation input ( Figure 3 The experimental results (Comparative Example 1) were confirmed when the compensation control input generation unit 15 in the middle was not present.

[0179] First, the experimental results of Comparative Example 1 are presented. Figure 11 .

[0180] in addition, Figure 11 The experimental results for (A) are a graph showing the response time between 3 and 24 seconds. Figure 11 (B) is Figure 11 The experimental results of (A) are plotted as a curve of the response in the initial step (between 3 and 6 seconds), and magnified to show the scale of the positional deviation. Furthermore, Figure 3 The parameters of the main control input generation unit 11 are set to Kp = 100 [N / m], Kd = 10 [N / m], K fu =12.5 [N] -1 ] .

[0181] like Figure 11 As shown, without the addition of compensation input, an overshoot of about 2 mm was generated, and a steady-state deviation of at least 0.05 mm remained.

[0182] Next, the experimental results of Example 1 are presented. Figure 12-13 .

[0183] in addition, Figure 12 The experimental results for (A) are a graph showing the response time between 3 and 6 seconds. Figure 12 (B) is Figure 12 The graph of the response in the initial step (between 3 and 6 seconds) of the experimental results in (A) is a magnified graph showing the scale of the position deviation. Furthermore, the parameters of the compensation control input generation unit 15 for generating force input to the target are set to K1 = 0.1 N, K2 = 2.0 N, and α = 0.8; the parameters of the compensation control input generation unit 15 for the control input are set to K1 = 20, K2 = 100, and α = 0.8; the deviation threshold is set to 20 nm; and the velocity threshold is set to 40 nm / ms. Other parameters are the same as in the comparative example.

[0184] like Figure 12 As shown, it can be confirmed that with the addition of a compensation input, the overshoot is significantly reduced compared to the case without a compensation input, and the deviation converges to below 50nm in steady state.

[0185] Figure 13 (A) magnification Figure 12 The range of (A) from 10.0s to 10.2s shows the change in the target force relative to the deviation.

[0186] like Figure 13 As shown in (A), a compensation input Fc is added to the main input Fr (in Figure 3The input after Yc is given to the force control unit, and it can be confirmed that as the deviation approaches 0, the compensation input Fc increases in the negative direction, thus acting as a strong brake. In addition, it can be confirmed that after overshoot, the compensation input Fc continues to take the maximum value. If the speed becomes 0, only the first control input Fr takes effect, and the displacement immediately moves towards the target value.

[0187] Figure 13 (B) shows the relationship with Figure 13 (A) The change in the control input relative to the deviation over the same time period. Additionally, the control input is represented by a 12-bit digital value, which corresponds to the voltage applied to the valve.

[0188] Add compensation input Uc to the main input U (in Figure 3 The input after Yc is given to the valve. Similar to the case where the target force is generated, it can be confirmed that the compensation input Uc acts as a strong brake, the overshoot is reduced, and it has the effect of immediately shifting towards the target value. Figure 13 (B)

[0189] The above results confirm that by incorporating the compensator assembly using the control input generation method of the present invention into the target force and control input, the target value can be positioned with high precision even when using a pneumatic actuator.

[0190] Furthermore, in the above examples, it was confirmed that by using a linear encoder with a resolution of 12.5 nm to measure the position of the worktable, the steady-state deviation can be kept below 50 nm. On the other hand, when using a compensator employing the control input generation method of the present invention, if the detection accuracy of the sensor is improved, a positioning accuracy corresponding to the detection accuracy of the sensor can be obtained.

[0191] The following confirms that by changing the sensor's position on the detection stage and using a compensator employing the control input generation method of this invention, a positioning accuracy corresponding to the sensor's detection accuracy can be obtained.

[0192] The experimental results are shown in Figure 23 , Figure 24 .

[0193] Figure 23 , Figure 24 The experimental results are shown when the same experiment as in Example 1 is performed using a linear encoder (Magnescale BL50H) with a resolution of 6.1 picometers as the sensor for detecting the position of the worktable.

[0194] in addition, Figure 23The experiment was conducted under the same conditions as in Example 1, except that a feedforward system without an external observer was used as the control system, which generated the main control input signal through PID control, a compensation input signal was added to the control system as the control input signal, and the aforementioned sensor was used.

[0195] also, Figure 24 Except for the use of the aforementioned sensor, it was implemented under the same conditions as in Example 1.

[0196] in addition, Figure 23 The experimental results for (A) are a graph showing the response from 0 to 24 seconds. Figure 23 (B) is Figure 23 The graph of the response in the initial step (0-4 seconds) of the experimental results in (A) is shown, and is a magnified view of part B. Similarly, Figure 24 The experimental results for (A) are a graph showing the response from 0 to 24 seconds. Figure 24 (B) is Figure 24 The graph of the response in the initial step (between 0 and 4 seconds) of the experimental results of (A) is shown in magnified form as part B.

[0197] First, such as Figure 23 As shown, it has been confirmed that even in a control system that generates the main control input signal through PID control, the steady-state deviation can be converged to below 40 nm simply by using a compensator employing the control input generation method of the present invention.

[0198] In addition, such as Figure 24 As shown, it was confirmed that even with a control system that is exactly the same as the control system in Example 1, by increasing the resolution of the sensor, the deviation in steady state can be converged to below ±5nm.

[0199] The above results confirm that by incorporating the compensator using the control input generation method of this invention into the target force and control input, high-precision positioning to the target value can be achieved even when using a pneumatic actuator, regardless of the control system. Furthermore, it has been confirmed that even with the same control system or the same compensator, changing the accuracy (resolution) of the sensor at the detection position can achieve positioning accuracy corresponding to the sensor's accuracy.

[0200] <Example 2> In Example 2, it was confirmed that even for minute changes in the target value, the target value can be located with high precision by incorporating a compensator using the control input generation method of the invention into the target generating force and control input.

[0201] In Example 2, the response (position deviation) was confirmed when a step input with a width of 0.1 μm was continuously applied every 3 seconds under steady-state conditions with a target step response of 50 mm. Furthermore, experiments were conducted under the same conditions as in Example 1, except that a step input with a width of 0.1 μm was continuously applied every 3 seconds.

[0202] The results are shown in Figure 14 .

[0203] like Figure 14 As shown, it can be confirmed that by continuously providing a step input with a width of 0.1 μm every 3 seconds, although a larger overshoot (around 1.5 μm) is generated than in Example 1, the deviation in steady state converges to below ±50 nm.

[0204] In addition, experimental results are shown when the same experiment as in Example 2 is performed, except that a linear encoder with a resolution of 6.1 picometers (Magnescale BL50H) is used for position detection on the worktable and a step input with a width of 1 nm is continuously given every 2 seconds.

[0205] The results are shown in Figure 25 .

[0206] In addition, Figure 25 In the middle, set the target value of 50mm to 0 on the vertical axis.

[0207] like Figure 25 As shown, it can be confirmed that even when a step input with a width of 1 nm is continuously given every 2 seconds, although there is a variation of about ±1 nm relative to the step input that becomes the target trajectory, it can roughly follow the target trajectory.

[0208] The above results confirm that by incorporating the compensator group using the control input generation method of the present invention into the target force and control input, even for minute changes in the target value, the pneumatic actuator can be used to precisely position the target value and accurately follow the changes in the target value.

[0209] <Example 3> In Example 3, it was confirmed that even when an external force is applied, by incorporating a compensator using the control input generation method of the present invention into the target force and control input, the target value can be located with high precision.

[0210] In Example 3, it was confirmed that... Figure 15The response (position deviation) was as shown in (A) with a 3 kg load suspended by a pulley and a step size of 0.1 μm. In addition, the same conditions as in Example 2 were applied, except that the 3 kg load was suspended and the step size was 0.1 μm.

[0211] The results are shown in Figure 15 (B)

[0212] Although the tensile force is approximately 50% of the maximum force calculated from the supply pressure and the pressure area of ​​the cylinder, it is still possible to position with the same accuracy as in the absence of tensile force under such conditions.

[0213] The above results confirm that by incorporating the compensator assembly using the control input generation method of the present invention into the target force and control input, even when an external force is applied, the pneumatic actuator can be used to position the target value with high precision.

[0214] <Example 4> In Example 4, it was confirmed that when the control system of the compensator using the control input generation method of the present invention is designed as a Type 2 control system with a disturbance observer, the target value can be located with high accuracy even when load changes occur during operation.

[0215] In Example 4, the load applied to the cylinder was varied as follows, and the movement state when the worktable was lifted to a predetermined position by the cylinder's action was confirmed. In the experiment, it was confirmed that before lifting the worktable to the predetermined position, (1) a load was applied midway through the movement from an unloaded state, (2) an additional load was applied midway through the movement from a state with a certain initial load, and (3) a first load was applied midway through the movement from an unloaded state, followed by an additional second load midway through the movement (see [reference]). Figure 18 (B)). In addition, except for extending and retracting the cylinder in the vertical direction and changing the applied load, the experiment was conducted under the same conditions as in Example 1.

[0216] Additionally, the no-load state refers to the load applied only to the plate located at the front end of the cylinder. Figure 18 The state of “m1” in (B). In addition, the load applied during the movement in experiment (1), the first load in experiment (3), and the initial load in experiment (2) are equivalent to Figure 18 The “m2” in (B). Furthermore, the additional load applied midway through the movement in experiment (2) and the second load in experiment (3) are equivalent to... Figure 18 The “m3” in (B).

[0217] In experiments (1) to (3), the apparatus is constructed as follows.

[0218] First, such as Figure 17 and Figure 18 As shown in (A), two worktables (worktable 1 and 2, each weighing 3 kg) supported by a frame are positioned at different heights along the movement path of the cylinder. These two worktables are positioned on the frame in a manner that allows them to move vertically but not downwards to a predetermined height, guided by the frame. The cylinder is positioned to contact these two worktables from below. That is, the cylinder is positioned so that its axis is parallel to the vertical direction, and it is positioned to extend and retract vertically. Furthermore, a plate is provided at the front end of the cylinder rod to stably support the worktables.

[0219] <Experiments (1) to (3)> The experiments (1) to (3) were carried out as follows.

[0220] In addition, in the experiment, such as Figure 19-21 As shown, the position control status of the cylinder was confirmed under the following conditions: First, in a Type 1 control system with a disturbance observer and a compensator using the control input generation method of the present invention is used (Conventional+FF, device 1); second, in a Type 2 control system with a disturbance observer and a compensator using the control input generation method of the present invention is used (Improved+FF, device 2). Both device 1 and device 2 implement feedforward (FF) control.

[0221] <Experiment (1)> In experiment (1), from the state of cylinder contraction ( Figure 18 The experiment begins in state (A). If the experiment begins, the cylinder is extended. Thus, the cylinder extends (in the unloaded state) under a load of m1 (the weight of the plate). Figure 18 During (B) X0 period), the plate contacts the worktable 1 and lifts the worktable 1. That is, at the setting position of the worktable 1, the weight m2 of the worktable 1 is applied as a load, and the load applied to the cylinder changes from m1 to m1+m2. If the cylinder extends further from the state of contact with the worktable 1, then the cylinder extends ( under the load of m1+m2) Figure 18 During (B) X1 period). And, the cylinder is retracted before the worktable 1 contacts the worktable 2. If the cylinder retracts, the worktable 1 is held in its original position by the frame, and the load applied to the cylinder changes. That is, if the cylinder retracts, the load corresponding to the weight m1 of the worktable 1 is reduced at the original position of the worktable 1, the load becomes the initial m1 state and retracts to the original length.

[0222] <Experiment (2)> In experiment (2), from the state of cylinder contraction ( Figure 18 The cylinder extends from the state of (A) where the plate is in contact with the workbench 1 and the load m2 of the workbench 1 is applied to the cylinder (i.e., the load m1+m2 is applied). If the experiment begins, the cylinder extends. Thus, the cylinder extends under a load of m1+m2. Figure 18 During (B) X1), workbench 1 contacts workbench 2 and workbench 2 is lifted. That is, at the setting position of workbench 2, the weight m3 of workbench 2 is applied as an additional load, and the load applied to the cylinder changes from m1+m2 to m1+m2+m3. Furthermore, if the cylinder extends further from the state where workbench 1 contacts workbench 2, the cylinder extends (under the state of load m1+m2+m3) Figure 18 (B) during X2 period). And, if the length is extended by a specified amount, the cylinder is retracted. If the cylinder retracts, the worktable 2 is held in its original position by the frame, and the load applied to the cylinder changes. That is, if the cylinder retracts, the load corresponding to the weight m3 of the worktable 2 is reduced in the original position of the worktable 2, the load becomes the initial m1+m2 state and retracts to the original length.

[0223] <Experiment (3)> In experiment (3), from the state of cylinder contraction ( Figure 18 The experiment begins in state (A). If the experiment begins, the cylinder is extended. Thus, the cylinder extends (under a load of m1, or in an unloaded state) Figure 18 During (B) X0 period), the initial plate contacts the worktable 1 below and lifts the worktable 1. That is, at the setting position of the worktable 1, the weight m2 of the worktable 1 is applied as a load, and the load applied to the cylinder changes from m1 to m1+m2. Then, if the cylinder extends further from the state of contact with the worktable 1, the cylinder extends during this period under the state of load m1+m2 ( Figure 18 During (B) X1 period). Shortly after, in a state where worktable 1 is in contact with worktable 2 and worktable 1 and worktable 2 are overlapping, the cylinder lifts worktable 1 and worktable 2. That is, at the setting position of worktable 2, the weight m3 of worktable 2 is further applied as a load, and the load applied to the cylinder changes from m1+m2 to m1+m2+m3. Furthermore, if the cylinder extends further from the state of contact with worktable 2, then during this period the cylinder extends (in the state of load m1+m2+m3). Figure 18During the X2 period of (B), if the cylinder extends by a specified amount, the cylinder extends and retracts, and worktables 1 and 2 are held in their original positions by the frame, and the load applied to the cylinder changes. That is, if the cylinder extends and retracts, the load corresponding to the weight m2 of worktable 2 is reduced at the original position of worktable 2, and the load corresponding to the weight m1 of worktable 1 is reduced at the original position of worktable 1, and the load becomes the initial m1 state and retracts to the original length.

[0224] The results are shown in Figure 19-21 .

[0225] <Experiments (1) to (3)> like Figure 19-21 As shown, in any of the experiments (1) to (3), it can be confirmed that in device 1, although there is some delay until the target value is reached when the load changes, it shows that it roughly follows the change of the target value.

[0226] On the other hand, in device 2, it can be confirmed that the delay relative to the target value is very small, showing a change that is almost equal to the target value.

[0227] The above results confirm that if a compensator employing the control input generation method of the present invention is used, the cylinder's movement can be controlled in a manner that exhibits good following performance relative to the target value, even under load variations. In particular, it has been confirmed that if the control system is designed as a Type 2 control system with a disturbance observer, the cylinder's movement can be controlled in a manner that approximately matches the target value, even under load variations.

[0228] Industrial applicability The control input generation device of the present invention is suitable as a device for generating control inputs for controlling the operation of an actuator that is actuated by air pressure.

[0229] Explanation of reference numerals in the attached figures 1 Control System 2 cylinders 3 piping 4. Air pressure source 5 Servo Valves 10. Control device 11 Main Control Input Generation Unit 15 Compensation Control Input Generation Unit 16 First control input generation function 17 Second control input generation function 17a Arrival Time Calculation Function 17b Predicted Steps Calculation Function 18 Output generation function S1 First control input S2 Second Control Input Yc Compensation Input L control quantity Lr target value e deviation.

Claims

1. A control input generation device, which generates control inputs for controlling a pneumatic actuator, characterized in that, have: The first control input generation function generates a first control input when there is a deviation between the control quantity and the target value. The second control input generation function generates a second control input that becomes larger as the deviation of the control quantity from the target value becomes smaller when there is a time change in the control quantity. as well as The output generation function generates the control input based on the first control input and the second control input.

2. The control input generation device as described in claim 1, characterized in that, The first control input generation function has the function of generating a constant steady-state control input as the first control input, or, It has the function of generating a control input as the first control input, which combines a variable control input that becomes smaller as the deviation of the control input from the target value becomes smaller, and a constant steady-state control input, when there is a time variation in the control input.

3. The control input generation device as described in claim 1, characterized in that, The second control input generation function calculates the number of prediction steps N until the control quantity matches the target value based on the deviation of the control quantity from the target value and the operating state of the pneumatic actuator, and generates the second control input based on the number of prediction steps N in a manner that the smaller the number of prediction steps N is, the larger it becomes.

4. The control input generation device as described in claim 2 or 3, characterized in that, The first control input generation function generates the first control input S1 based on Equation 1. The second control input generation function generates the second control input S2 based on Equation 2. The output generation function generates the control input based on Equation 3. Formula 1 S1 = K1·sgn(e) Formula 2 S2=K2 / N α ·sgn(v) Formula 3 Yc=K1·sgn(e)-K2 / N α ·sgn(v) e: Deviation of the control quantity from the target value v: Displacement speed of the pneumatic actuator α: A power exponent used to adjust the timing at which the suppressive force for inhibiting pneumatic actuator movement begins to take effect. K1 = Gain in generating propulsion K2 = Gain that produces the inhibitory force N = Number of prediction steps.

5. The control input generation device as described in claim 4, characterized in that, The first control input generation function is based on Equation 2-1 to set the gain K1 for compensating for the nonlinear friction of the pneumatic actuator and generating propulsion force. The second control input generation function is based on Equation 2-2 to set the gain K2 that reduces the suppression of overshoot relative to the target value. Equation 2-1 The case where K1 = Ka|e| > Eth K1 = Ka / (Eth·|e|) (case where |e|≤Eth) Ka: Specified control input Eth: Specified displacement |e|: The absolute value of the deviation e Equation 2-2 K2=Kb|v|>Vth K2=0|v|≤Vth Kb: Specified control input Vth: Specified displacement velocity |v|: The absolute value of velocity v.

6. The control input generation device as described in claim 3, characterized in that, The second control input generation function has: The arrival time calculation function calculates the time Ta required for the control quantity to reach the target value based on the deviation of the control quantity from the target value. as well as The prediction step calculation function calculates the prediction step number N by dividing the time Ta by the sampling period Δt of the deviation of the detection control quantity from the target value. The prediction step calculation function sets the prediction step number N as a threshold if the prediction step number N is less than a specified threshold.

7. The control input generation device as described in claim 6, characterized in that, The arrival time calculation function calculates the arrival time Ta based on Equation 4. Formula 4 (1 / 2)·a·Ta 2 +v·Ta-e=0 a: Displacement acceleration of the pneumatic actuator v: Displacement speed of the pneumatic actuator Ta: Arrival time, i.e., arrival time - current time. e: Deviation, i.e., target position - current position.

8. The control input generation device as described in claim 3, characterized in that, The second control input generation function calculates the prediction step N based on the following Equation 5. Formula 5 N = e / ΔL = e / (v·Δt) e: Deviation of the control quantity from the target value v: Displacement speed of the pneumatic actuator △L: The change in displacement since one sampling time.

9. The control input generation device as described in claim 1, characterized in that, The target value is the intended position of the object that is moved by the pneumatic actuator, or the driving speed of the pneumatic actuator, or the pressure generated by the pneumatic actuator.

10. A control device for controlling a pneumatic actuator, characterized in that, It has the control input generation device as described in claim 1.

11. A control device for controlling a pneumatic actuator, characterized in that, have: The main control input generation unit generates the main control input for controlling the pneumatic actuator; and The compensation control input generation unit generates compensation control inputs that correct the main control inputs. The compensation control input generation unit is the control input generation device as described in claim 1.

12. The control device as described in claim 10 or 11, characterized in that, The control device is a type 2 control system with a disturbance observer.

13. The control device as described in claim 12, characterized in that, The control device implements feedforward control.

14. An actuator, characterized in that, have: A moving part that uses air pressure to operate; and The control unit that controls the movement of the motion unit. The control unit is the control device as described in claim 10 or 11.

15. An operational circuit, which takes a target value and a control quantity as input signals, characterized in that, have: The first control input generation unit generates a first control input when there is a deviation between the control quantity and the target value. The second control input generation unit generates a second control input that becomes larger as the deviation of the control quantity from the target value becomes smaller when there is a time change in the control quantity. as well as The output generation unit generates and outputs control inputs based on the first control input and the second control input.

16. A control board having input terminals for inputting target values ​​and control quantities as input signals, and output terminals for generating and outputting control inputs, characterized in that, The control panel is equipped with: The main controller generates main control inputs based on signals input from the input terminals; The compensator generates a compensation control input based on the signal input from the input terminal; as well as The output generation unit is an adder circuit that combines the main control input and the compensation control input to generate a control input and supplies it to the output terminal. The compensator has the following characteristics: The first control input generation function generates a first control input when there is a deviation between the control quantity and the target value. The second control input generation function generates a second control input that becomes larger as the deviation of the control quantity from the target value increases, given the time-varying nature of the control quantity. The output generation function generates and outputs control inputs based on the first control input and the second control input.

17. A control board having input terminals for inputting target values ​​and control quantities as input signals, and output terminals for generating and outputting control inputs, characterized in that, The control panel is equipped with: The compensator generates a compensation control input based on a signal input from the input terminal; and The output generation unit is an adder circuit that combines the compensation control input and the externally supplied control input to generate a control input and supplies it to the output terminal. The compensator has the following characteristics: The first control input generation function generates a first control input when there is a deviation between the control quantity and the target value. The second control input generation function generates a second control input that becomes larger as the deviation of the control quantity from the target value increases, given the time-varying nature of the control quantity. The output generation function generates and outputs control inputs based on the first control input and the second control input.

18. A method for generating control inputs, characterized in that, The first control input is generated when there is a deviation between the control quantity and the target value. When there is a time-varying control variable, the second control input becomes larger as the deviation of the generated control variable from the target value decreases. The control input is generated based on the first control input and the second control input.

19. The control input generation method as described in claim 18, characterized in that, The first control input is a constant steady-state control input. or, When there is a time variation in the control quantity, the first control input is a combination of a variable control input that becomes smaller as the deviation of the control quantity from the target value becomes smaller and a constant steady-state control input.

20. The control input generation method as described in claim 18, characterized in that, Based on the deviation of the control quantity from the target value and the operating state of the pneumatic actuator, the prediction step number N until the control quantity matches the target value is calculated, and based on the prediction step number N, the second control input is generated in such a way that the smaller the prediction step number N is, the larger it becomes.

21. The control input generation method as described in claim 18 or 19, characterized in that, The first control input S1 is generated based on Equation 1. The second control input S2 is generated based on Equation 2. The control input is generated based on Equation 3. Formula 1 S1 = K1·sgn(e) Formula 2 S2=K2 / N α ·sgn(v) Formula 3 Yc=K1·sgn(e)-K2 / N α ·sgn(v) e: Deviation of the control quantity from the target value v: Displacement speed of the pneumatic actuator α: A power exponent used to adjust the timing at which the suppressive force for inhibiting pneumatic actuator movement begins to take effect. K1 = Gain in generating propulsion K2 = Gain that produces the inhibitory force N = Number of prediction steps.

22. The control input generation method as described in claim 21, characterized in that, Based on Equation 2-1, the gain K1 for compensating for the nonlinear friction of the pneumatic actuator and generating propulsion force is set. Based on Equation 2-2, the gain K2 is set to generate the suppression force that reduces overshoot relative to the target value. Equation 2-1 The case where K1 = Ka|e| > Eth The case where K1 = Ka / (Eth·|e|)|e|≤Eth Ka: Specified control input Eth: Specified displacement |e|: The absolute value of the deviation e Equation 2-2 K2=Kb|v|>Vth K2=0|v|≤Vth Kb: Specified control input Vth: Specified displacement velocity |v|: The absolute value of velocity v.

23. The control input generation method as described in claim 20, characterized in that, Based on the deviation of the control quantity from the target value, the time Ta required for the control quantity to reach the target value is calculated. The prediction step number N = Ta / Δt is obtained by dividing the time Ta by the sampling period Δt of the deviation of the detection control quantity from the target value. If the number of prediction steps N is less than a predetermined threshold, the number of prediction steps N is set as the threshold.

24. The control input generation method as described in claim 23, characterized in that, The arrival time Ta is calculated based on Equation 4. Formula 4 (1 / 2)·a·Ta 2 +v·Ta-e=0 a: Displacement acceleration of the pneumatic actuator v: Displacement speed of the pneumatic actuator Ta: Arrival time, i.e., arrival time - current time. e: Deviation, i.e., target position - current position.

25. The control input generation device as described in claim 20, characterized in that, The second control input generation function calculates the prediction step N based on the following Equation 5. Formula 5 N = e / ΔL = e / (v·Δt) e: Deviation of the control quantity from the target value v: Displacement speed of the pneumatic actuator △L: The change in displacement since one sampling time.

26. The control input generation method as described in claim 18, characterized in that, The target value is the intended position of the object that is moved by the pneumatic actuator, or the driving speed of the pneumatic actuator, or the pressure generated by the pneumatic actuator.

27. A control input generation program, characterized in that, It is a procedure for a control device that controls the action of a pneumatic actuator to execute the control input generation method according to any one of claims 18 to 25.

28. A storage medium storing a control input generation program, characterized in that, It is a storage medium storing the control input generation program as described in claim 27.

Images