Parameter adjustment device, positioning system, positioning method, and manufacturing method for electronic component packaging base

Abstract

To obtain a parameter adjustment device capable of reducing a possibility that a motor will be operated by using a parameter which is undesirable for a user even during a time period until a parameter adjustment is completed.SOLUTION: A parameter adjustment device 10 for outputting an operation-purpose command parameter to a positioning control device 20, which comprises: an operation-purpose command parameter candidate generating unit 11 that generates a plurality of operation-purpose command parameter candidates; an advance data acquisition unit 12 that acquires advance data including a feature amount when performing a trial positioning operation; a feature amount model generating unit 13 that generates a trained feature amount model by using the advance data; a feature amount prediction unit 14 that uses the trained feature amount model to output a feature amount prediction result corresponding to each operation-purpose command parameter candidate; and an operation-purpose command parameter determining unit 15 that selects one of a plurality of operation-purpose command parameter candidates based on the feature amount prediction result.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a parameter adjustment device that adjusts parameters used in positioning control, a positioning system, a positioning method, and a method for manufacturing a board mounting electronic components. [Background technology]

[0002] In devices that repeatedly move a head, which is a moving part to be controlled, by driving a motor, such as electronic component mounting machines and semiconductor manufacturing equipment, high-speed positioning control of the motor is required to improve productivity. However, when a motor operates at high speed, mechanical vibrations may occur. In such cases, if the command shape of the motor position command is appropriately adjusted, high-speed positioning control can be achieved even under conditions affected by mechanical vibrations. Therefore, it is required to appropriately adjust the command shape of the position command.

[0003] Patent Document 1 discloses a positioning control device for adjusting a command shape. This positioning control device learns the relationship between a position command parameter used to generate a position command and an evaluation value, and determines the position command parameter based on the learning result. Specifically, the positioning control device of Patent Document 1 adjusts the parameters so that the obtained optimal parameters satisfy the constraint conditions by giving a penalty when the magnitude of the residual vibration exceeds an allowable value. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2020 / 075316 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the above-mentioned conventional technology has a problem in that there is a high possibility that the motor will operate using parameters that are not desirable for the user during the period until adjustment to appropriate parameters is completed.

[0006] The present disclosure has been made in consideration of the above, and aims to provide a parameter adjustment device that can reduce the possibility of a motor operating using parameters that are not desirable for the user, even during the period until parameter adjustment is completed. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems and achieve the object, the parameter adjustment device disclosed herein outputs driving command parameters to a positioning control device having a command waveform generation unit that generates a position command, which is a time-series signal that commands the position of a motor that performs a positioning operation to move a movable part a target moving distance, based on driving command parameters, and a drive control unit that supplies power based on the position command to drive the motor, the parameter adjustment device having the command waveform generation unit that generates a plurality of driving command parameter candidates that are candidates for the driving command parameters, and a pre-data acquisition unit that acquires pre-data including feature quantities indicative of the state of at least one of the positioning control device, the movable part, and the motor when a trial positioning operation is performed. The parameter adjustment device further includes a feature model generation unit that uses prior data to generate a learned feature model for inferring features from driving command parameters, a feature prediction unit that inputs a plurality of driving command parameter candidates to the learned feature model and outputs predicted values ​​of features output for each of the plurality of driving command parameter candidates as feature prediction results, and a driving command parameter determination unit that selects one from the plurality of driving command parameter candidates based on the feature prediction result and determines it as a driving command parameter to be used in generating a position command. Effect of the Invention

[0008] According to the present disclosure, it is possible to reduce the possibility that the motor will operate using parameters that are not desirable for the user, even during the period until parameter adjustment is completed. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating an example of a configuration of a positioning system according to a first embodiment. [Diagram 2] FIG. 13 is a diagram showing an example of waveforms of a position command, a velocity command, an acceleration command, and a jerk used in the first embodiment; [Diagram 3] FIG. 1 is an explanatory diagram of an example of a method for calculating a feature amount in the first embodiment; [Figure 4] FIG. 4 is an explanatory diagram of an example of the operation of the driving command parameter determination unit when the feature quantity shown in FIG. 3 is used; [Diagram 5] FIG. 1 is a diagram showing an example of a neural network used by a feature model generation unit. [Figure 6] 1 is a flowchart illustrating the operation of the positioning system. [Figure 7] FIG. 1 is a diagram showing dedicated hardware for implementing the functions of a parameter adjustment device according to a first embodiment. [Figure 8] FIG. 1 shows a configuration of a control circuit for implementing the functions of a parameter adjustment device according to a first embodiment. [Figure 9] Flowchart for explaining a method for manufacturing an electronic component mounting board using a positioning system [Figure 10] FIG. 1 shows a configuration of a positioning system according to a second embodiment. [Figure 11] Calculation method for settling time of positioning operation [Figure 12] 11 is a flowchart for explaining the operation of the positioning system according to the second embodiment. [Figure 13] A flowchart for explaining the details of step S203 in FIG. [Figure 14] FIG. 13 is a diagram showing an example of a prediction result of an evaluation value and a feature amount using a Gaussian process regression model according to the second embodiment. [Figure 15] FIG. 13 is a diagram showing a configuration of a positioning system according to a third embodiment. [Figure 16] 11 is a flowchart for explaining the operation of a parameter adjustment device according to a third embodiment. [Figure 17] A flowchart for explaining the details of step S302 in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A parameter adjustment device, a positioning system, a positioning method, and a manufacturing method for an electronic component mounting board according to embodiments of the present disclosure will be described in detail below with reference to the drawings.

[0011] Embodiment 1 1 is a diagram illustrating an example of a configuration of a positioning system 1 according to a first embodiment. The positioning system 1 includes a parameter adjustment device 10, a positioning control device 20, a motor 30, a movable unit 40, and a sensor 50. The movable unit 40 is movably installed on a stand (not shown), and the motor 30 and the movable unit 40 are part of a positioning device including the stand.

[0012] The parameter adjustment device 10 has a function of adjusting the values ​​of operation command parameters, which are parameters used for positioning control. The parameter adjustment device 10 outputs the adjusted operation command parameters to the positioning control device 20. The positioning control device 20 is a device that controls a positioning operation for moving the movable part 40 by a target moving distance. The positioning control device 20 has a function of supplying a current to the motor 30 to drive the motor 30.

[0013] The motor 30 applies torque or force to the movable part 40 via a ball screw to move the movable part 40. The type of the motor 30 is not limited as long as it can drive the movable part 40. For example, the motor 30 is a rotary servo motor, a linear motor, a stepping motor, or the like. Here, the position of the motor 30 is a position corresponding to the operation of the motor 30, and is a position corresponding to the amount of movement of the mover of the motor 30. For example, when the motor 30 is a rotary servo motor, the rotation start position is set as the origin, and the rotation angle of the motor 30 from this origin position is multiplied by a proportional coefficient to give a positive or negative sign according to the rotation direction, and the sum of the values ​​added from the start of rotation to a certain point in time can be set as the position of the motor 30 at that point in time. In addition, the speed of the motor 30 can be set as a value obtained by differentiating the position of the motor 30 once with respect to time, and the acceleration of the motor 30 can be set as a value obtained by differentiating the position of the motor 30 twice with respect to time.

[0014] The movable part 40 is moved by the motor 30 by a desired target moving distance. The movable part 40 is a machine or a part that requires positioning control. The movable part 40 is, for example, a head part of an electronic component mounting machine or a semiconductor manufacturing device.

[0015] The sensor 50 detects the position of the motor 30 or the position of the movable part 40, and outputs the detected position detection value as a sensor signal to the parameter adjustment device 10A. The position detection value is the result of detection by the sensor 50. The sensor 50 is, for example, a rotary encoder, a linear scale, or a laser displacement meter.

[0016] The positioning control device 20 executes positioning control based on operating conditions input from outside the positioning control device 20 and operating command parameters output by the parameter adjustment device 10. The operating conditions are information including a target movement distance for driving the motor 30. The target movement distance is a desired distance for the movable part 40 to move. The positioning control device 20 executes positioning control for the movable part 40 so as to satisfy the operating conditions.

[0017] The positioning control device 20 generates a command waveform, which is the shape of a position command for driving the motor 30 to move the movable part 40 by a target moving distance, based on command parameters that define the shape of the position command. The position command is a time-series signal that commands the position of the motor 30. The acceleration shapes in the acceleration section and the deceleration section are determined based on the command parameters.

[0018] The positioning control device 20 has a command waveform generating unit 21 and a drive control unit 22. The command waveform generating unit 21 generates a command waveform, which is the shape of a position command used for positioning control, based on operating conditions input from outside the positioning control device 20 and operating command parameters output by the parameter adjustment device 10. The operating command parameters are command parameters received by the command waveform generating unit 21 to perform positioning operation, among command parameters that define the command waveform of a position command. The command waveform generating unit 21 outputs the generated command waveform to the drive control unit 22.

[0019] 2 is a diagram showing an example of the waveforms of the position command, velocity command, acceleration command, and jerk used in the first embodiment. The velocity command is the first-order differential of the position command, the acceleration command is the second-order differential of the position command, and the jerk is the first-order differential of the acceleration command, that is, the jerk. Fig. 2 shows an example of the shapes of each command and jerk when operating conditions including information indicating a target travel distance are provided to the positioning control device 20.

[0020] As shown in FIG. 2, the acceleration command in the first embodiment is a command that indicates a trapezoidal shape in the acceleration direction from the first section to the third section, is constant in the fourth section, and is a command that indicates a trapezoidal shape in the deceleration direction from the fifth section to the seventh section. The first section is the section where acceleration starts, the third section is the section where acceleration ends, the fifth section is the section where deceleration starts, and the seventh section is the section where deceleration basically ends. The time length of the mth section is defined as the mth time length Tm, where m is an integer from 1 to 7. For example, the time length of the first section is the first time length T1.

[0021] In the acceleration command in FIG. 2, the trapezoidal shape of the acceleration section from the first section to the third section and the trapezoidal shape of the deceleration section from the fifth section to the seventh section may not be congruent. In FIG. 2, the acceleration section is trapezoidal, but the first time length T1 and the third time length T3 of the acceleration section may be set to 0, and the shape of the acceleration command may be rectangular. In the first embodiment, seven parameters from the first time length T1 to the seventh time length T7 are command parameters. The command waveform is specified based on the command parameters and the target moving distance. A method of calculating the command waveform will be described later.

[0022] Returning to the description of FIG. 1, the drive control unit 22 drives the motor 30 by supplying power to the motor 30 so that the position of the motor 30 follows the position command. For example, the drive control unit 22 acquires information indicating the position of the motor 30, calculates a value of a current to be supplied to the motor 30 based on PID (Proportional-Integral-Differential) control so that the deviation between the position of the motor 30 and the position command is small, and supplies power of the calculated value of the current to the motor 30. Here, the information indicating the position of the motor 30 can be acquired using, for example, a rotary encoder, a linear scale, a laser displacement meter, or the like. Note that the drive control unit 22 may use any control method as long as it causes the position of the motor 30 to follow the position command. For example, the drive control unit 22 may perform two-degree-of-freedom control by adding feedforward control to feedback control. In addition, the drive control unit 22 drives the motor 30 so that the position of the motor 30 follows the position command here, but may detect the position of the movable part 40 as a signal for feedback control and drive the motor 30 so that the position of the movable part 40 follows the position command.

[0023] Here, a method for calculating a command waveform will be described. The command waveform generating unit 21 calculates a position command based on input operation command parameters. For the sake of explanation, the example shown in FIG. 2 will be referred to again.

[0024] The magnitude of acceleration in the second section is set as Aa, and the magnitude of acceleration in the sixth section is set as Ad. Note that the magnitude of acceleration in the second section, Aa, and the magnitude of acceleration in the sixth section, Ad, are dependent variables of the command parameters, so there is no freedom in setting them.

[0025] In the first section, where time t is greater than or equal to 0 and less than T1, the acceleration command A1(t) of the first section at time t can be calculated using the following formula (1), the velocity command V1(t) of the first section at time t can be calculated using the following formula (2), and the position command P1(t) of the first section at time t can be calculated using the following formula (3).

[0026]

number

[0027]

number

[0028]

number

[0029] Furthermore, in the second section where time t is greater than or equal to T1 and less than T1+T2, the acceleration command A2(t) of the second section at time t can be calculated using the following formula (4), the velocity command V2(t) of the second section at time t can be calculated using the following formula (5), and the position command P2(t) of the second section at time t can be calculated using the following formula (6).

[0030]

number

[0031]

number

[0032]

number

[0033] Furthermore, in the third section, where time t is greater than or equal to T1+T2 and less than T1+T2+T3, the acceleration command A3(t) of the third section at time t can be calculated using the following formula (7), the velocity command V3(t) of the third section at time t can be calculated using the following formula (8), and the position command P3(t) of the third section at time t can be calculated using the following formula (9).

[0034]

number

[0035]

number

[0036]

number

[0037] Furthermore, in the fourth section, where time t is greater than or equal to T1+T2+T3 and less than T1+T2+T3+T4, the acceleration command A4(t) of the fourth section at time t can be calculated using the following formula (10), the velocity command V4(t) of the fourth section at time t can be calculated using the following formula (11), and the position command P4(t) of the fourth section at time t can be calculated using the following formula (12).

[0038]

number

[0039]

number

[0040]

number

[0041] Furthermore, in the fifth section, where time t is greater than or equal to T1+T2+T3+T4 and less than T1+T2+T3+T4+T5, the acceleration command A5(t) of the fifth section at time t can be calculated using the following formula (13), the velocity command V5(t) of the fifth section at time t can be calculated using the following formula (14), and the position command P5(t) of the fifth section at time t can be calculated using the following formula (15).

[0042]

number

[0043]

number

[0044]

number

[0045] Furthermore, in the 6th section, where time t is greater than or equal to T1+T2+T3+T4+T5 and less than T1+T2+T3+T4+T5+T6, the acceleration command A6(t) of the 6th section at time t can be calculated using the following formula (16), the velocity command V6(t) of the 6th section at time t can be calculated using the following formula (17), and the position command P6(t) of the 6th section at time t can be calculated using the following formula (18).

[0046]

number

[0047]

number

[0048]

number

[0049] Furthermore, in the 7th section, where time t is greater than or equal to T1+T2+T3+T4+T5+T6 and less than T1+T2+T3+T4+T5+T6+T7, the acceleration command A7(t) of the 7th section at time t can be calculated using the following formula (19), the velocity command V7(t) of the 7th section at time t can be calculated using the following formula (20), and the position command P7(t) of the 7th section at time t can be calculated using the following formula (21).

[0050]

number

[0051]

number

[0052]

number

[0053] At the terminal time t=T1+T2+T3+T4+T5+T6+T7, the speed command needs to match 0, and the position command needs to match the target moving distance D. Therefore, at the terminal time t=T1+T2+T3+T4+T5+T6+T7, the following formulas (22) and (23) hold.

[0054]

number

[0055]

number

[0056] These two equations determine the magnitude of acceleration Aa in the second section and the magnitude of acceleration Ad in the sixth section.

[0057] By the above procedure, the command waveform of the position command can be calculated based on the command parameters and the target movement distance D.

[0058] In this case, since the acceleration is a linear function of time in the first, third, fifth, and seventh sections, the jerk, which is the first derivative of the acceleration, is a non-zero constant value in these sections. In other words, the first time length T1, the third time length T3, the fifth time length T5, and the seventh time length T7 can be said to determine the time during which the jerk is a non-zero constant value. Here, a non-zero constant value means a constant value greater than 0 or less than 0.

[0059] In these sections, a parameter may be used to specify the magnitude of the jerk instead of the duration. For example, if the magnitude of the jerk in the first section is J1, the relationship between the first duration T1 and the jerk J1 is expressed by the following formula (24).

[0060]

number

[0061] In other words, determining the time period during which the jerk is a non-zero constant value as a parameter is equivalent to determining the magnitude of the jerk during which the jerk is a non-zero constant value as a parameter. In this way, the method of determining the parameters that define the command waveform is arbitrary and is not limited to the method described in this embodiment.

[0062] Returning to the explanation of Fig. 1, the parameter adjusting device 10 includes an operational command parameter candidate generating unit 11, a prior data acquiring unit 12, a feature quantity model generating unit 13, a feature quantity predicting unit 14, an operational command parameter determining unit 15, an evaluation value determining unit 17, and a feature quantity determining unit 18.

[0063] The driving command parameter candidate generating unit 11 generates a plurality of driving command parameter candidates that are candidates for driving command parameters. The driving command parameter candidate generating unit 11 outputs the generated plurality of driving parameter candidates to the feature amount predicting unit 14. The driving command parameter candidate generating unit 11 can generate the driving command parameter candidates, for example, based on random numbers. Minimum and maximum values ​​that each of the driving command parameters can take are set based on the driving conditions. Between the set minimum and maximum values, the driving command parameter candidate generating unit 11 generates a plurality of driving command parameters based on random numbers.

[0064] However, the method by which the driving command parameter candidate generating unit 11 generates driving parameter candidates is not limited to the method using random numbers. For example, the driving command parameter candidate generating unit 11 may generate driving parameter candidates based on a grid. In this case, the driving command parameter candidate generating unit 11 sets the interval width between the minimum value and the maximum value, and provides a grid within the interval in which each parameter can be taken. Then, the driving command parameter candidate generating unit 11 refers to the parameters of the grid in ascending or descending order of value, and sets them as driving command parameter candidates. The driving command parameter candidate generating unit 11 may also generate driving command parameter candidates using a method such as random sampling or Latin hypermodular method.

[0065] The prior data acquisition unit 12 acquires prior data including feature quantities indicating the state of at least one of the positioning control device 20, the movable part 40, and the motor 30 when a trial positioning operation is performed. The prior data acquisition unit 12 can acquire, as prior data, a set of command parameters used when the trial positioning operation is performed and the above feature quantities when the trial positioning operation is performed using the command parameters. The prior data acquisition unit 12 outputs the acquired prior data to the feature model generation unit 13.

[0066] Here, the feature amounts acquired as the advance data include measurement values ​​of the states of the positioning control device 20, the motor 30, the movable part 40, etc., and command values ​​to the motor 30. For example, the feature amounts acquired as the advance data include signals acquired by sensors indicating the position, speed, acceleration, temperature, current, torque, etc. of the motor 30, signals acquired by sensors indicating the position, speed, acceleration, temperature, etc. of the movable part 40, and signals acquired by sensors indicating the temperature, sound emitted, power consumption, installation state of the stand, etc. of the positioning control device 20.

[0067] The feature quantity model generation unit 13 generates a trained feature quantity model for inferring features from driving command parameters, using the prior data output by the prior data acquisition unit 12. The feature quantity model generation unit 13 outputs the generated trained feature quantity model to the feature quantity prediction unit 14.

[0068] The feature prediction unit 14 inputs each of the multiple driving command parameter candidates into the learned feature model, and outputs the predicted value of the feature output from the learned feature model for each of the multiple driving command parameter candidates to the driving command parameter determination unit 15 as a feature prediction result.

[0069] The driving command parameter determination unit 15 selects one from among the multiple driving command parameter candidates based on the feature prediction result output by the feature prediction unit 14, and determines it as the driving command parameter to be used for generating a position command. The driving command parameter determination unit 15 outputs the determined driving command parameter to the positioning control device 20.

[0070] The evaluation value determination unit 17 has a function of calculating an evaluation value corresponding to the operation command parameter based on the sensor signal output by the sensor 50. The evaluation value is a value for evaluating the positioning operation. The evaluation value can be expressed, for example, by the settling time of the positioning operation. If the evaluation value is Q and the settling time of the positioning operation is Tst, the evaluation value Q can be expressed as "Q = -Tst". Note that the calculation method of the evaluation value Q shown here is an example. When calculating the evaluation value Q based on the settling time Tst, the evaluation value Q should be larger as the settling time Tst is shorter. In addition, any value other than the settling time Tst that can evaluate the positioning operation may be used. For example, the evaluation value may include a term representing a penalty of a feature amount described later. The evaluation value determination unit 17 outputs the calculated evaluation value to the operation command parameter determination unit 15.

[0071] When a positioning operation is performed using the driving command parameters, the feature determination unit 18 calculates a feature based on the sensor signal output by the sensor 50, associates the calculated feature with the driving command parameters, and outputs the result to the driving command parameter determination unit 15.

[0072] In adjusting the operation command parameters for positioning control, if a change occurs in the characteristics of the controlled object or in the installation environment after the adjustment is completed, the operation command parameters must be readjusted. For example, when the installation location is moved to a different floor of a factory building, the frequency characteristics of the natural vibration mode in which the installation stiffness is dominant change due to the change in the stiffness of the floor surface. Due to this change in characteristics, the operation command parameters must be readjusted every time the installation location is changed. For this reason, in this embodiment, when the operation command parameters are readjusted, a feature quantity model generated based on prior data is used, making it possible to predict the feature quantities when each of the operation command parameter candidates is used.

[0073] As described above, because a change in the situation during operation of the positioning system 1 requires readjustment of the command parameters, it is desirable for the prior data acquisition unit 12 to hold prior data as different data before and after the change in the situation during operation of the positioning system 1. For example, when the change in the situation during operation is a change in the installation location, the prior data acquisition unit 12 holds prior data for each installation location. For example, the prior data obtained when the positioning operation is performed with the positioning system 1 installed at the first installation location can be set as the first prior data, and the prior data obtained when the positioning operation is performed after the positioning system 1 is moved to the second installation location can be set as the second prior data. In addition, by regenerating the feature model in association with the readjustment of the operation command parameters, the accuracy of the feature prediction result of the feature model can be improved.

[0074] Here, a change in the installation location is given as an example of a change in the situation of the positioning system 1, but the present embodiment is not limited to such an example. For example, the change in situation may be an event that can change the installation situation or mechanical characteristics of the positioning system 1, such as shipping of the positioning system 1, changes in the environment such as the temperature and humidity around the positioning system 1, or performing installation work again at the same installation location.

[0075] In addition, when the advance data acquisition unit 12 holds advance data for each state of the positioning system 1, it is desirable to hold data indicating the state of the positioning system 1 when the advance data was acquired in association with the feature amount. For example, the data indicating the state of the positioning system 1 when the advance data was acquired includes at least one of information indicating the state of fixing the positioning device having a stand on which the movable part 40 is installed to the floor surface, the atmospheric temperature around the positioning device, the vibration state of the floor surface on which the positioning device is installed, the usage time of the positioning device, and the state of the lubricant of the positioning device. Here, the state of the lubricant is the physical properties, pressure, temperature, remaining amount, usage time, etc. of the lubricant. As a result, the state of the positioning system 1 itself and the state of the environment in which the positioning system 1 is installed are included in the advance data, and these states are reflected in the feature amount model and the feature amount that is the output of the feature amount model, thereby improving the adjustment accuracy of the operation command parameters.

[0076] The prior data acquiring unit 12 may acquire prior data from outside the parameter adjusting device 10 in a format in which driving command parameters corresponding to the trial positioning operation are associated with feature quantities when the trial positioning operation is executed, or may have a function as a feature quantity determining unit that acquires detection values ​​of a sensor or the like when the trial positioning operation is executed from outside the parameter adjusting device 10 and determines feature quantities. In this case, the feature quantity determining unit can determine feature quantities for driving command parameters corresponding to the trial positioning operation based on the result of comparing the position command with the sensor signal.

[0077] FIG. 3 is an explanatory diagram of an example of a method for calculating a feature amount in the first embodiment. The horizontal axis of FIG. 3 indicates time, and the vertical axis indicates the magnitude of deviation. FIG. 3 shows the change over time of the deviation between the target moving distance and the detected value of the moving distance of the movable part 40 when positioning control is performed using a command waveform generated based on the operation command parameters. Here, IMP is the allowable value of the deviation in the positioning control, and is a predetermined value. AMP is the peak value of the residual vibration after the magnitude of the deviation becomes smaller than the allowable value IMP, and is called the residual vibration maximum amplitude. Based on the allowable value IMP and the residual vibration maximum amplitude AMP, the feature amount determination unit can calculate the feature amount using the following formula (25).

[0078]

number

[0079] C in the formula (25) is an example of a feature value, and is a value obtained by subtracting the maximum residual vibration amplitude AMP from the allowable value IMP. When the feature value C is a value of 0 or more, the amplitude of the residual vibration falls within the allowable value IMP.

[0080] FIG. 4 is an explanatory diagram of an example of the operation of the driving command parameter determination unit 15 when the feature quantity shown in FIG. 3 is used. FIG. 4 shows the time response of the deviation of the positioning control when driving command parameters corresponding to two patterns of command waveforms are used. One of the two driving command parameters is called the first driving command parameter, and the other is called the second driving command parameter. The upper diagram of FIG. 4 shows the time response of the deviation of the position of the movable part 40 when the positioning control is performed using the first driving command parameter, and the lower diagram of FIG. 4 shows the time response of the deviation of the moving distance of the movable part 40 when the positioning control is performed using the second driving command parameter.

[0081] In the case of the first operation command parameter, the deviation becomes smaller than the allowable value IMP at time Tst600, and thereafter exceeds the allowable value IMP by the time it reaches the residual vibration maximum amplitude AMP602, which is the peak of the residual vibration, at time Tst601. That is, at time Tst601, the characteristic amount C is a value of 0 or less. In the case of the second operation command parameter, the deviation becomes smaller than the allowable value IMP at time Tst610, and at subsequent times, the deviation does not exceed the allowable value IMP. That is, even when it reaches the residual vibration maximum amplitude AMP602, which is the peak of the residual vibration, at time Tst612, the residual vibration maximum amplitude AMP602 is smaller than the allowable value IMP. In this case, the characteristic amount C is a value greater than 0.

[0082] When two patterns of driving command parameters as shown in FIG. 4 are candidates for driving command parameters, the driving command parameter determination unit 15 can select a second driving command parameter whose value of the feature quantity C is greater than 0, based on the value of the feature quantity C.

[0083] The method of calculating the characteristic amount is not limited to the above. For example, the characteristic amount may be calculated as the maximum or minimum value of the acceleration of the command waveform, the maximum or minimum value of the speed of the command waveform, the maximum or minimum value of the torque of the motor 30, the time required for the vibration of the movable part 40 to damp after the positioning is completed, the volume of the sound generated by the positioning system 1 during operation, or the like.

[0084] Here, the feature quantity model will be described in detail. The feature quantity model is a model that expresses the relationship between command parameters and feature quantities. The feature quantity model is, for example, a neural network model that uses command parameters as input and features as output. The neural network model acts as a function Nn that predicts a feature quantity predicted value Cest1 corresponding to an input command parameter Pcmd, as shown in the following formula (26).

[0085]

number

[0086] The feature model generating unit 13 generates a trained feature model based on the prior data. In this embodiment, supervised learning is performed on the neural network model constituting the feature model, and the trained model is output as the trained feature model. Here, supervised learning refers to a model in which a large amount of data sets of input and result, i.e., labels, are provided to a learning device, and the features in the data set are learned, and a result is estimated from the input. Furthermore, the feature model generating unit 13 updates the feature model based on the command parameters and the feature every time a positioning operation based on the command parameters is performed, and outputs the model as the trained feature model.

[0087] A neural network is composed of an input layer made up of a plurality of neurons, an intermediate layer made up of a plurality of neurons, and an output layer made up of a plurality of neurons. The intermediate layer is also called a hidden layer. The intermediate layer may be one layer or two or more layers. FIG. 5 is a diagram showing an example of a neural network used by the feature model generating unit 13. For example, in the case of a three-layered neural network as shown in FIG. 5, when a plurality of inputs are input to the input layer (X1 to X7), the values ​​are multiplied by weights W1 (W111 to W179) and input to the intermediate layer (Y1 to Y9), and the results are further multiplied by weights W2 (W211 to W291) and output from the output layer (Z1). This output result changes depending on the values ​​of the weights W1 and W2.

[0088] In this embodiment, the neural network learns the relationship between the command parameters and the feature quantities by so-called supervised learning in accordance with a data set created based on a combination of the driving command parameters and the feature quantities.

[0089] That is, the neural network learns by inputting driving command parameters to the input layer and adjusting the weights W1 and W2 so that the results output from the output layer approach the feature quantities.

[0090] The feature model generating unit 13 may learn the feature model according to a data set created for a plurality of positioning systems 1. The feature model generating unit 13 may acquire data sets from a plurality of positioning systems 1 used at the same site, or may learn the feature model using data sets collected from a plurality of positioning systems 1 operating independently at different sites. Furthermore, the positioning system 1 that collects the data set may be added to the targets midway, or conversely, may be removed from the targets midway. A machine learning device that has learned a feature model for a certain positioning system 1 may be attached to another positioning system 1.

[0091] In the above embodiment, supervised learning using a neural network is applied as the learning algorithm of the feature model, but the learning algorithm of the feature model is not limited to this method. For example, deep learning that learns to extract the feature itself can be used, or machine learning can be performed according to other known methods, such as genetic programming, inductive logic programming, and support vector machines.

[0092] Here, the operation of the positioning system 1 will be described. Fig. 6 is a flowchart for explaining the operation of the positioning system 1. First, in the positioning system 1, the prior data acquisition unit 12 of the parameter adjustment device 10 acquires prior data (step S100) and outputs the acquired prior data to the feature model generation unit 13. The feature model generation unit 13 performs a learning process for generating a feature model using the prior data output by the prior data acquisition unit 12 (step S101), and outputs the learned feature model to the feature prediction unit 14.

[0093] The driving command parameter candidate generating unit 11 sets driving command parameter candidates (step S102). The feature predicting unit 14 inputs the set driving command parameter candidates to the learned feature model output by the feature model generating unit 13, acquires the feature output from the learned feature model, and outputs it to the driving command parameter determining unit 15, and the driving command parameter determining unit 15 judges whether or not the constraint condition is predicted to be satisfied based on the output feature (step S103). If the constraint condition is predicted not to be satisfied (step S103: No), the parameter adjusting device 10 repeats the process from step S102. If the constraint condition is predicted to be satisfied (step S103: Yes), the driving command parameter determining unit 15 sets the set driving command parameter candidates as driving command parameters of the adjustment result (step S104), drives the movable part 40 by the motor 30 using the set driving command parameters, and receives a sensor signal output by the sensor 50 at that time (step S105). The evaluation value determination unit 17 determines an evaluation value for the driving command parameter based on the received sensor signal (step S106). The feature amount determination unit 18 determines a feature amount based on the received sensor signal (step S107). The feature amount model generation unit 13 regenerates a feature amount model using the determined feature amount and the corresponding driving command parameter (step S108). The driving command parameter determination unit 15 determines whether or not a termination condition is satisfied (step S109). If the termination condition is not satisfied (step S109: No), the driving command parameter determination unit 15 saves the command parameter and the evaluation value, and returns to the process of step S102. If the termination condition is satisfied (step S109: Yes), the driving command parameter determination unit 15 determines an adjustment result (step S111) and ends the process.

[0094] Next, a hardware configuration of the parameter adjustment device 10 according to the first embodiment will be described. Each functional unit of the parameter adjustment device 10 is realized by a processing circuit. These processing circuits may be realized by dedicated hardware, or may be a control circuit using a CPU (Central Processing Unit).

[0095] When the above processing circuits are realized by dedicated hardware, they are realized by a processing circuit 90 shown in Fig. 7. Fig. 7 is a diagram showing dedicated hardware for realizing the functions of the parameter adjustment device 10 according to the first embodiment. The processing circuit 90 is a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination of these.

[0096] When the above processing circuit is realized by a control circuit using a CPU, the control circuit is, for example, a control circuit 91 having a configuration shown in Fig. 8. Fig. 8 is a diagram showing the configuration of the control circuit 91 for realizing the functions of the parameter adjustment device 10 according to the first embodiment. As shown in Fig. 8, the control circuit 91 includes a processor 92 and a memory 93. The processor 92 is a CPU, and is also called a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, a DSP (Digital Signal Processor), or the like. The memory 93 is, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable ROM), or an EEPROM (Electrically EPROM), a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD (Digital Versatile Disk).

[0097] When the above processing circuit is realized by the control circuit 91, it is realized by the processor 92 reading and executing a program corresponding to the processing of each component, which is stored in the memory 93. The memory 93 is also used as a temporary memory for each process executed by the processor 92. The program executed by the processor 92 may be provided in a state stored in a storage medium, or may be provided via a communication path such as the Internet.

[0098] In the above, the type of the movable part 40 is not particularly limited, but the positioning system 1 can be applied to, for example, a manufacturing apparatus for an electronic component mounting board that mounts electronic components on the electronic component mounting board. In this case, the movable part 40 is, for example, a gripping part that grips an electronic component. A method for manufacturing an electronic component mounting board when the positioning system 1 is applied to the manufacturing apparatus for an electronic component mounting board will be described.

[0099] 9 is a flowchart for explaining a method for manufacturing a board mounted with electronic components using positioning system 1. The method for manufacturing a board mounted with electronic components includes a gripping step (step S20) of gripping an electronic component with movable part 40, which is a gripping part, a positioning step (step S21) of moving movable part 40, which is a gripping part, by a target movement distance while holding the electronic component, and an arrangement step (step S22) of releasing the electronic component at the position where movable part 40 stops, thereby arranging the electronic component on the mounting board.

[0100] As described above, according to the first embodiment, it is possible to provide a parameter adjustment device 10 that outputs operation command parameters to the positioning control device 20. The positioning control device 20 has a command waveform generation unit 21 that generates a position command, which is a time-series signal that commands the position of the motor 30 that executes a positioning operation to move the movable part 40 by a target moving distance, based on the operation command parameters, and a drive control unit 22 that supplies power based on the position command to drive the motor 30. The parameter adjustment device 10 has an operation command parameter candidate generation unit 11 that generates a plurality of operation command parameter candidates that are candidates for the operation command parameters, and a prior data acquisition unit 12 that acquires prior data including a feature amount that indicates a state of at least one of the positioning control device 20, the movable part 40, and the motor 30 when a trial positioning operation is executed. The parameter adjustment device 10 also includes a feature model generation unit 13 that uses prior data to generate a learned feature model for inferring features from driving command parameters, a feature prediction unit 14 that inputs a plurality of driving command parameter candidates to the learned feature model and outputs a predicted value of a feature output for each of the plurality of driving command parameter candidates as a feature prediction result, and a driving command parameter determination unit 15 that selects one from the plurality of driving command parameter candidates based on the feature prediction result and determines it as a driving command parameter to be used for generating a position command. With the above configuration, a search for command parameters is performed based on prior data, so that it is possible to reduce the possibility that the positioning device will operate with parameters that are undesirable to the user during the period until adjustment to appropriate parameters is completed.

[0101] The advance data acquisition unit 12 can associate at least one of the motor 30, the movable unit 40, and the state of the positioning device, which includes a stand on which the movable unit 40 is movably installed, with a feature quantity indicating the state of at least one of the positioning control device 20, the movable unit 40, and the motor 30, and store the associated advance data as advance data, and store the advance data acquired by installing the positioning device at a first installation location as the first advance data, and store the advance data acquired by installing the positioning device at a second installation location different from the first installation location as the second advance data. As described above, the advance data includes data indicating the state of the positioning device and the state of the surroundings of the positioning device, and the like, so that the state of the positioning device and the state of the surroundings of the positioning device can be reflected in the learned feature model and the feature quantity output from the learned feature model. Therefore, the operation command parameter determination unit 15 can accurately determine the operation command parameters.

[0102] Furthermore, according to the first embodiment, it is possible to provide a positioning system 1 including a positioning control device 20 having a motor 30 that performs a positioning operation to move the movable part 40 by a target moving distance, a command waveform generating unit 21 that generates a position command, which is a time-series signal that commands the position of the motor 30, based on operation command parameters, and a drive control unit 22 that supplies power based on the position command to drive the motor 30, and a parameter adjusting device 10 that outputs the operation command parameters to the positioning control device 20.

[0103] Moreover, according to the first embodiment, it is also possible to provide a positioning method using the positioning system 1. The positioning method includes a step in which the parameter adjustment device 10 generates a plurality of driving command parameter candidates which are candidates for driving command parameters to be output to the positioning control device 20, a step in which the parameter adjustment device 10 acquires advance data including features indicating a state of at least one of the positioning control device 20, the movable part 40, and the motor 30 when the motor 30 trial-executes a positioning operation in which the movable part 40 is moved by a target moving distance, a step in which the parameter adjustment device 10 uses the advance data to generate a learned feature model for inferring the features from the driving command parameters, and a step in which the parameter adjustment device 10 inputs the plurality of driving command parameter candidates into the learned feature model. The method includes the steps of: outputting predicted values ​​of feature quantities output for each of a plurality of driving command parameter candidates as feature quantity prediction results; a step in which the parameter adjusting device 10 selects one from the plurality of driving command parameter candidates based on the feature quantity prediction result and determines it as an driving command parameter to be used for generating a position command; a step in which the positioning control device 20 generates a position command, which is a time-series signal that commands the position of the motor 30, based on the driving command parameters output by the parameter adjusting device 10; a step in which the positioning control device 20 supplies power based on the position command to drive the motor 30; and a step in which the motor 30 executes a positioning operation to move the movable part 40 by a target moving distance.

[0104] Moreover, according to the first embodiment, it is also possible to provide a method for manufacturing a board mounted with electronic components. The method for manufacturing a board mounted with electronic components includes a gripping step in which a gripper, which is movable unit 40, grips an electronic component, a positioning step in which the gripper, while gripping the electronic component, is moved a target movement distance on the mounting board, and an arrangement step in which the gripper releases the electronic component at a position where the gripper stops, thereby arranging the electronic component on the mounting board. Specifically, the positioning step includes multiple steps of the positioning method described above.

[0105] Embodiment 2 In the second embodiment, a configuration using Gaussian process regression instead of a neural network will be described.

[0106] 10 is a diagram showing a configuration of a positioning system 1A according to the second embodiment. The positioning system 1A has a parameter adjustment device 10A, a positioning control device 20, a motor 30, a movable part 40, and a sensor 50. The positioning system 1A has a parameter adjustment device 10A instead of the parameter adjustment device 10 of the positioning system 1 according to the first embodiment. The following mainly describes the parts that differ from the first embodiment.

[0107] Parameter adjustment device 10A has an operational command parameter candidate generation unit 11, a preliminary data acquisition unit 12A, a feature model generation unit 13, a feature prediction unit 14A, an operational command parameter determination unit 15A, an evaluation value model generation unit 16, an evaluation value determination unit 17A, and a feature determination unit 18.

[0108] The prior data acquiring unit 12A acquires data including a plurality of sets of command parameters, evaluation values, and feature quantities as prior data. As with the prior data in the first embodiment, a situation may change between when the prior data is acquired and when the positioning system 1A in the present embodiment is in operation.

[0109] The feature quantity prediction unit 14A inputs the driving command parameter candidates to the learned feature quantity model, and outputs the driving command parameter candidates in addition to the prediction result of the feature quantity output from the learned feature quantity model. At this time, the feature quantity prediction unit 14A outputs the driving command parameter candidates to each of the driving command parameter determination unit 15A and the evaluation value determination unit 17A.

[0110] The evaluation value determination unit 17A has a function of calculating an evaluation value corresponding to the driving command parameter based on the sensor signal output by the sensor 50, and a function of estimating an evaluation value corresponding to the driving command parameter candidate using the learned evaluation value model generated by the evaluation value model generation unit 16. The evaluation value is a value for evaluating the positioning operation. The evaluation value can be expressed, for example, by the settling time of the positioning operation. If the evaluation value is Q and the settling time of the positioning operation is Tst, the evaluation value Q can be expressed as "Q = -Tst". Note that the calculation method of the evaluation value Q shown here is an example. When calculating the evaluation value Q based on the settling time Tst, the evaluation value Q should be larger as the settling time Tst is shorter. In addition, any value other than the settling time Tst that can evaluate the positioning operation may be used. For example, the evaluation value may include a term representing a penalty of a feature amount described later.

[0111] Here, a method for calculating the settling time of the positioning operation based on the sensor signal will be described. Fig. 11 is an explanatory diagram of the method for calculating the settling time of the positioning operation. Fig. 11 shows the time response of the deviation between the target moving distance and the detection value of the actual moving distance when positioning control is performed using each command waveform generated based on the first to third sets of operation command parameters. In Fig. 11, (a) shows the time response of the deviation when the first set of operation command parameters is used, (b) shows the second set of operation command parameters, and (c) shows the third set of operation command parameters.

[0112] For example, the settling time Tst is the time from the start of positioning control to the completion of positioning when the deviation between the target travel distance and the detected travel distance becomes smaller than the allowable value IMP. The settling time Tst obtained from the sensor signal when the positioning operation is performed using the first set of operation command parameters Pcmd1 is set to settling time Tst1, and the evaluation value Q calculated from this settling time Tst1 is set to evaluation value Q1. Similarly, the settling time Tst obtained from the sensor signal when the positioning operation is performed using the second set of operation command parameters Pcmd2 is set to settling time Tst2, and the evaluation value Q calculated from this settling time Tst2 is set to evaluation value Q2. Moreover, the settling time Tst obtained from the sensor signal when the positioning operation is performed using the third set of operation command parameters Pcmd3 is set to settling time Tst3, and the evaluation value Q calculated from this settling time Tst3 is set to evaluation value Q3. At this time, the evaluation values ​​corresponding to the three sets of operation command parameters are calculated as "Q1=-Tst1", "Q2=-Tst2", and "Q3=-Tst3".

[0113] First, when the parameter adjustment device 10A outputs the first set of operation command parameters Pcmd1, positioning control is executed using a position command generated using the first set of operation command parameters Pcmd1. The evaluation value determination unit 17A acquires a settling time Tst1 corresponding to the first set of operation command parameters Pcmd1 based on the position indicated by the sensor signal output by the sensor 50 at this time. Here, the settling time is defined as the time from the start of positioning control to the completion of positioning at which the magnitude of the deviation between the target travel distance and the detected travel distance becomes smaller than the allowable value IMP. The evaluation value Q1 corresponding to the first set of operation command parameters is calculated as "Q1 = -Tst1". Here, in the example of FIG. 11, Tst2 <Tst1<Tst3であるため、Q2> Q1>Q3, and the evaluation value Q2 is the largest.

[0114] The evaluation value determination unit 17A associates an evaluation value calculated based on the driving command parameters and the sensor signal with the driving command parameters, and outputs the associated evaluation value to the evaluation value model generation unit 16. Furthermore, the evaluation value determination unit 17A inputs driving command parameter candidates to a learned evaluation value model, and outputs an evaluation value output from the learned evaluation value model to the driving command parameter determination unit 15A, in association with the driving command parameter candidates.

[0115] The evaluation value model generation unit 16 generates an evaluation value model for inferring an evaluation value from driving command parameter candidates based on the prior data output by the prior data acquisition unit 12A, and outputs the generated evaluation value model to the evaluation value determination unit 17A as a learned evaluation value model. In addition, the evaluation value model generation unit 16 can regenerate the evaluation value model using the driving command parameters output by the evaluation value determination unit 17A and the evaluation values ​​associated with the driving command parameters.

[0116] The feature determination unit 18 calculates a feature based on a sensor signal acquired when a positioning operation is performed using the driving command parameters, and outputs the calculated feature to the driving command parameter determination unit 15A in association with the driving command parameters.

[0117] The driving command parameter determination unit 15A selects one driving command parameter candidate from among the multiple driving command parameter candidates based on the feature prediction result and the evaluation value, and determines it as the driving command parameter to be used by the positioning control device 20.

[0118] 12 is a flowchart for explaining the operation of the positioning system 1A according to the second embodiment. First, the prior data acquisition unit 12A acquires prior data (step S200). Then, the evaluation value model generation unit 16 generates an evaluation value model based on the prior data (step S201). In addition, the feature quantity model generation unit 13 generates a feature quantity model based on the prior data (step S202).

[0119] The parameter adjustment device 10A determines the next search point, that is, the driving command parameters for the next positioning operation (step S203). The determination method will be described later.

[0120] The parameter adjusting device 10A sets the determined operation command parameters in the positioning control device 20 to perform a positioning operation (step S204). The sensor 50 receives a sensor signal that is a time-series signal of the position of the movable part 40 (step S205).

[0121] The evaluation value determination unit 17A determines an evaluation value based on the sensor signal acquired in the positioning operation (step S206). The evaluation value determination unit 17A holds the determined evaluation value. The feature amount determination unit 18 determines a feature amount based on the sensor signal acquired in the positioning operation (step S207). The feature amount determination unit 18 holds the determined feature amount.

[0122] The parameter adjustment device 10A judges whether a predetermined termination condition is satisfied (step S208). As a specific example, the parameter adjustment device 10A judges that the termination condition is satisfied when the number of times the positioning operation is performed exceeds a predetermined upper limit. The termination condition in step S208 is not limited to the above example. For example, the parameter adjustment device 10A may use a termination condition based on an evaluation value or a feature amount of the positioning operation. For example, the parameter adjustment device 10A may judge that the termination condition is satisfied when the best value of the evaluation value of the positioning operation is not updated a predetermined number of times or more. The parameter adjustment device 10A may also judge that the termination condition is satisfied when the number of consecutive times that the update width of the operation command parameter falls below a predetermined width exceeds a predetermined number of times.

[0123] If the termination condition is not satisfied (step S208: No), the evaluation value model generation unit 16 regenerates the evaluation value model based on the evaluation value held by the evaluation value determination unit 17A (step S209) and updates the learned evaluation value model. In addition, the feature amount model generation unit 13 regenerates the feature amount model based on the feature amount held by the feature amount determination unit 18 (step S210) and updates the learned feature amount model. After performing the process of step S210, the parameter adjustment device 10A returns to the process of step S203.

[0124] When the termination condition is satisfied (step S208: Yes), the driving command parameter determination unit 15A determines the adjustment result based on the evaluation value and the feature amount (step S211). For example, the driving command parameter determination unit 15A can select, from the evaluation values ​​held in the evaluation value determination unit 17A, those whose corresponding feature amount exceeds a predetermined threshold value, and determine, as the adjustment result, the driving command parameter associated with the best evaluation value among the selected ones.

[0125] Fig. 13 is a flowchart for explaining the details of step S203 in Fig. 12. First, the driving command parameter candidate generating unit 11 generates driving command parameter candidates (step S220). The feature amount predicting unit 14A inputs the driving command parameter candidates into a learned feature amount model and acquires the feature amount output as a predicted value of the feature amount (step S221). At this time, the feature amount predicting unit 14A may calculate the value taking into consideration factors such as standard deviation and variance for defining a confidence interval of the predicted value, as described later.

[0126] The driving command parameter determination unit 15A compares the predicted value of the feature with a threshold value set for the feature, and judges whether or not the constraint condition is predicted to be satisfied when the driving command parameter candidate is used (step S222). A label of the prediction result may be assigned to each driving command parameter candidate, and the label may be output as a constraint prediction result. As an example of a method of assigning labels to the prediction results, a label of "1" may be assigned to a feature predicted to be highly likely to satisfy the constraint condition, and a label of "0" may be assigned to a feature not predicted to be highly likely to satisfy the constraint condition.

[0127] If it is predicted that the constraint condition is satisfied (step S222: Yes), the evaluation value determination unit 17A then inputs the driving command parameter candidate into the learned evaluation value model to calculate a predicted value of the evaluation value (step S223). At this time, as described later, the value may be calculated taking into consideration factors such as standard deviation and variance for defining a confidence interval of the predicted value. The driving command parameter determination unit 15A judges whether a predetermined termination condition is satisfied (step S224). If the termination condition is satisfied (step S224: Yes), the driving command parameter determination unit 15A determines the driving command parameter to be set (step S225) and ends the process. The termination condition may be, for example, when a driving command parameter candidate is obtained that is predicted to obtain an evaluation value equal to or better than the best value of the evaluation value held by the evaluation value determination unit 17A. Here, the driving command parameter determination unit 15A can select the driving command parameter candidate that has the best predicted value of the evaluation value among the ones stored so far as the driving command parameter for obtaining the next evaluation value and feature amount.

[0128] If it is not predicted that the constraint condition is satisfied (step S222: No), the operation command parameter determination unit 15A judges whether or not the termination condition is satisfied (step S226), and if the termination condition is satisfied (step S226: Yes), the process is terminated, and if the termination condition is not satisfied (step S226: No), the process is repeated from step S220. The termination condition used here may be, for example, the number of consecutive times that the constraint condition is not satisfied exceeds a predetermined value. Depending on the conditions of the machine device and the external setting values, there may be an extremely small number of command parameters that satisfy the constraint condition, and this is the termination condition for this reason.

[0129] Fig. 14 is a diagram showing an example of prediction results of an evaluation value and a feature amount using a Gaussian process regression model according to the second embodiment. (a1) of Fig. 14 is a graph showing a relationship between a command parameter Pcmd and an evaluation value Q estimated by a Gaussian process regression model in a state where three pairs of a command parameter and an evaluation value corresponding to the command parameter are obtained. Note that, although in the first embodiment, the command parameter is a set of multiple setting values, in Fig. 14, for the sake of explanation, it is assumed that there is one setting value.

[0130] FIG. 14(a2) is a graph showing a case where one more pair of a command parameter and an evaluation value is obtained from the state of FIG. 14(a1) by the operation of parameter adjustment device 10A.

[0131] FIG. 14(b1) is a graph showing the relationship of the feature quantity C to the command parameter Pcmd estimated by a Gaussian process regression model in a state where three pairs of command parameters and feature quantities corresponding to the command parameters are obtained.

[0132] FIG. 14(b2) is a graph showing a case where one more pair of command parameter and feature quantity is obtained from the state of FIG. 14(b1) by the operation of parameter adjustment device 10A.

[0133] Each graph in Fig. 14 depicts the expected value, which is the prediction result of the Gaussian process regression model, the confidence interval, the upper limit of the confidence interval, the lower limit of the confidence interval, and a threshold value previously set for the feature amount. The dashed-dotted line indicates the expected value, the hatched area indicates the confidence interval, the solid line indicates the upper and lower limits of the confidence interval, and the dashed line indicates the threshold value. In addition, three sets of data already obtained at the points (a1) and (b1) in Fig. 14 are indicated by black circles, and one set of data further obtained by the operation of the parameter adjustment device 10A is indicated by a white circle.

[0134] In FIG. 14(a1), as a function of the command parameter Pcmd, the expected value of the evaluation value is Qe1(Pcmd), the upper limit of the confidence interval is Qu1(Pcmd), and the lower limit of the confidence interval is Ql1(Pcmd).

[0135] In (b1) of FIG. 14, as a function of the command parameter Pcmd, the expected value of the feature is Ce1(Pcmd), the upper limit of the confidence interval is Cu1(Pcmd), and the lower limit of the confidence interval is Cl1(Pcmd).

[0136] In FIG. 14(a2), as a function of the command parameter Pcmd, the expected value of the evaluation value is Qe2(Pcmd), the upper limit of the confidence interval is Qu2(Pcmd), and the lower limit of the confidence interval is Ql2(Pcmd).

[0137] In (b2) of FIG. 14, as a function of the command parameter Pcmd, the expected value of the feature is Ce2(Pcmd), the upper limit of the confidence interval is Cu2(Pcmd), and the lower limit of the confidence interval is Cl2(Pcmd).

[0138] 14(a1) and (b1), the minimum command parameter for which the lower limit Cl1(Pcmd) of the confidence interval of the feature exceeds the threshold is PcmdLSB1, and the maximum command parameter is PcmdRSB1. The command parameters of the set that have already been obtained are Pcmd21, Pcmd22, and Pcmd23.

[0139] Also, in (a2) and (b2) of FIG. 14, the minimum command parameter at which the lower limit value Cl2(Pcmd) of the confidence interval of the feature amount exceeds the threshold value is defined as PcmdLSB2, and the maximum command parameter is defined as PcmdRSB2.

[0140] Regarding the set of command parameters Pcmd21, Pcmd22, and Pcmd23 that have already been obtained at the time of (a1) and (b1) of FIG. 14, they have the same values in (a2) and (b2) of FIG. 14.

[0141] Also, in (a2) and (b2) of FIG. 14, the values of the newly acquired set of command parameters are defined as PcmdNew.

[0142] Hereinafter, a method for determining the command parameter PcmdNew for newly acquiring a set by the parameter adjustment device 10A will be described starting from the states of (a1) and (b1) of FIG. 14. Referring to (b1) of FIG. 14, the command parameter for which the feature amount is likely to exceed the threshold value, that is, the command parameter that is likely to satisfy the constraint condition, can be said to be the command parameter between PcmdLSB1 and PcmdRSB1 at which the lower limit value Cl1(Pcmd) of the confidence interval of the feature amount exceeds the threshold value. That is, when the command parameter Pcmd satisfies "Thre2 < Cl1(Pcmd)", it can be determined that the constraint condition is likely to be satisfied.

[0143] In the situation shown in FIG. 14, a simple method for searching for a command parameter with the largest possible evaluation value while keeping the possibility of violating the constraints low is to select a command parameter that is expected to improve the evaluation value between PcmdLSB1 and PcmdRSB1. For example, a command parameter with the highest upper limit Qu1(Pcmd) of the confidence interval is selected. In (a1) of FIG. 14, the upper limit of the evaluation value in the command parameter PcmdLSB1 is set to Qu1(PcmdRSB1). As is clear from FIG. 14, in this example, the command parameter with the largest upper limit Qu1(Pcmd) of the evaluation value between PcmdLSB1 and PcmdRSB1 where the lower limit Cl1(Pcmd) of the confidence interval of the feature exceeds the threshold value is PcmdRSB1. In this way, the command parameter with the largest upper limit of the confidence interval of the evaluation value within the range where the constraints are likely to be satisfied may be set as the command parameter PcmdNew for acquiring a new set.

[0144] In the second embodiment, as a method for searching for a command parameter with the highest upper limit Cu1(Pcmd) of the confidence interval of the evaluation value, a plurality of command parameters that are likely to satisfy the constraint conditions are extracted from a plurality of operation command parameter candidates generated by the operation command parameter candidate generation unit 11. Furthermore, the value of the upper limit Qu1(Pcmd) of the confidence interval of the evaluation value for the extracted command parameters is calculated, and the command parameter with the highest upper limit Qu1(Pcmd) of the confidence interval of the evaluation value is set as the command parameter PcmdNew for acquiring a new set.

[0145] In addition, in order to determine the command parameters PcmdNew for obtaining a new set, other methods such as the steepest descent method, conjugate gradient method, Newton method, quasi-Newton method, Gauss-Newton method, interior point method, Nelder-Mead method, Powell method, CMA-ES method, Latin Hypercube method, particle swarm optimization method, etc. may be used as a method for searching for the command parameters that maximize the upper limit Cu1(Pcmd) of the confidence interval of the evaluation value, in addition to the above.

[0146] In this example, the interval predicted to be highly likely to satisfy the constraint conditions for the command parameters is a continuous interval from PcmdLSB to PcmdRSB, but it may be divided into multiple intervals. Whether this interval is a continuous interval or divided into multiple intervals depends on the value of the threshold value Thre2 and the shape of the lower limit value Cl1(Pcmd) of the confidence interval of the feature. Even in such a case, for a set of command parameters that are highly likely to satisfy the constraint conditions, the command parameter with the largest upper limit value of the confidence interval of the evaluation value can be selected as the command parameter PcmdNew for obtaining a new set.

[0147] In this example, the command parameter PcmdNew for obtaining a new set is the maximum command parameter PcmdRSB1 at which the lower limit Cl1(Pcmd) of the confidence interval of the feature exceeds a threshold value, but the value of the command parameter PcmdNew for obtaining a new set is not limited to the above example. While the determination of the command parameter PcmdNew for obtaining a new set is repeated, the upper limit Cu1(Pcmd) of the confidence interval of the evaluation value may become maximum at a point approximately in the middle of the section from PcmdLSB1 to PcmdRSB1. If the upper limit Cu1(Pcmd) of the confidence interval of the evaluation value becomes maximum at a point approximately in the middle of the section from PcmdLSB1 to PcmdRSB1, the point approximately in the middle of the section from PcmdLSB1 to PcmdRSB1 is selected as the command parameter PcmdNew for obtaining a new set. In the following description, the word "point" may be used to mean "command parameter for operation" or "command parameter".

[0148] In the above example, the upper limit Qul(Pcmd) of the confidence interval of the evaluation value is referenced to determine the command parameter PcmdNew for obtaining a new set. Alternatively, the expected value Qe1(Pcmd) or the lower limit Ql1(Pcmd) of the evaluation value may be used. When the expected value Qe1(Pcmd) is used, the command parameter with the largest expected value Qe1(Pcmd) of the evaluation value can be selected as the command parameter PcmdNew for obtaining a new set from among the command parameters that are likely to satisfy the constraint conditions. When the lower limit Ql1(Pcmd) is used, the command parameter with the least possibility of the evaluation value being small can be selected as the command parameter PcmdNew for obtaining a new set from among the command parameters that are likely to satisfy the constraint conditions.

[0149] In the above example, the lower limit Cl1(Pcmd) of the confidence interval of the feature is referenced to determine the set of command parameters that are likely to satisfy the constraint conditions. Alternatively, the expected value Ce2(Pcmd) of the feature or the upper limit Cu1(Pcmd) of the confidence interval may be used. By using the expected value Ce2(Pcmd) of the feature or the upper limit Cu1(Pcmd) of the confidence interval to determine the set of command parameters that are likely to satisfy the constraint conditions, the range of command parameters that can be determined to be likely to satisfy the constraint conditions can be widened. This makes it possible to select a command parameter PcmdNew for obtaining a new set from a wider range of command parameter sets, thereby obtaining a better evaluation value.

[0150] In the above example, the standard deviation of the Gaussian distribution is σ, and the range of ±1σ is set as the confidence interval of the predicted value, but the method of defining the confidence interval is not limited to this. In an ideal Gaussian distribution, the elements included in the range of ±1σ are approximately 68%, the elements included in the range of ±2σ are approximately 95%, and the elements included in the range of ±3σ are approximately 99.7%. Therefore, it can be said that the newly obtained pair is more likely to be included in the range of ±2σ than the range of ±1σ, and more likely to be included in the range of ±3σ than the range of ±2σ. However, the wider the range, the narrower the range of command parameters that can be predicted to be more likely to satisfy the constraint conditions, and global search becomes more difficult. The range of the confidence interval used for prediction is preferably about ±1σ to ±3σ. In addition, the set value of the width of this confidence interval may be set separately for those used for predicting the evaluation value and those used for predicting the feature amount.

[0151] Also, as a configuration different from the present embodiment, the parameter adjustment device 10A may be configured to easily expand the range of command parameters that are likely to satisfy the constraint conditions. Specifically, the feature model generation unit 13 regenerates the feature model by giving the upper limit value of the feature confidence interval to the command parameter candidate that is determined to satisfy the constraint conditions within the prediction range of the current Gaussian process regression model. The operation command parameter candidates predicted to be likely to satisfy the constraint conditions based on the regenerated feature model are set as an expanded set (Expander). Furthermore, among the operation command parameter candidates predicted to be likely to satisfy the constraint conditions, a region of command parameters in which the upper limit value of the confidence interval of the evaluation value exceeds the maximum value of the lower limit of the confidence interval of the evaluation value is set as a maximizing set (Maximaizer). In the union of the expanded set and the maximizing set, the width of the confidence interval of the evaluation value and the width of the confidence interval of the feature are calculated, and the operation command parameter candidate with the largest width is set as the operation command parameter.

[0152] The operation of parameter adjustment device 10A described above makes it possible to perform a parameter search that takes into account the reliability of evaluation values ​​and predicted values ​​of feature quantities, thereby enabling an efficient search while avoiding parameter operations that are undesirable to the user.

[0153] As described above, the parameter adjustment device 10A according to the second embodiment can further include an evaluation value determination unit 17A and an evaluation value model generation unit 16 in addition to the configuration of the parameter adjustment device 10 according to the first embodiment. The evaluation value determination unit 17A can determine an evaluation value of a positioning operation corresponding to an operation command parameter of a trial positioning operation based on a sensor signal that detects the position of the movable part 40 or the motor 30 when the trial positioning operation using the operation command parameter is performed, and can determine an evaluation value corresponding to an operation command parameter candidate using a learned evaluation value model for inferring an evaluation value from the operation command parameter candidate. In addition, the evaluation value model generation unit 16 can generate a learned evaluation value model for inferring an evaluation value from the operation command parameter candidate using prior data. Furthermore, in the second embodiment, the operation command parameter determination unit 15A can select one from a plurality of operation command parameter candidates based on the feature amount prediction result and the evaluation value, and determine it as an operation command parameter to be used by the positioning control device 20. By determining the driving command parameters based on the evaluation values, it is possible to preferentially search for driving command parameter candidates with high evaluation values, thereby shortening the time required to determine the driving command parameters.

[0154] Furthermore, the parameter adjustment device 10A according to the second embodiment can further include a feature amount determiner 18 that determines a feature amount for an operation command parameter corresponding to a trial positioning operation, in addition to the configuration of the parameter adjustment device 10 according to the first embodiment. The feature amount determiner 18 can determine the feature amount based on a result of comparing the position command with a sensor signal that detects the position of the movable part 40 or the motor 30. This makes it possible to avoid a load on the positioning control device 20 that occurs due to an excessive difference between the position command and the actual position.

[0155] Moreover, according to the second embodiment, it is also possible to provide a positioning method using the positioning system 1A, similar to the first embodiment. Although detailed description is omitted in the second embodiment, it is also possible to provide a method for manufacturing a board mounted with electronic components by applying the positioning system 1A to an apparatus for manufacturing a board mounted with electronic components, similar to the first embodiment.

[0156] Embodiment 3 In the third embodiment, a parameter adjustment method is described which performs an efficient search while satisfying the constraint conditions by preferentially searching for command parameters predicted to have high evaluation values ​​among those predicted to satisfy the constraint conditions.

[0157] 15 is a diagram showing the configuration of a positioning system 1B according to the third embodiment. The positioning system 1B includes a parameter adjustment device 10B, a positioning control device 20, a motor 30, a movable part 40, and a sensor 50. The following mainly describes the parts that differ from the second embodiment.

[0158] Parameter adjustment device 10B has an operating command parameter candidate generation unit 11B, a pre-data acquisition unit 12, a feature model generation unit 13, a feature prediction unit 14B, an operating command parameter determination unit 15B, an evaluation value model generation unit 16, an evaluation value determination unit 17B, and a feature determination unit 18.

[0159] The driving command parameter candidate generating unit 11B generates driving command parameter candidates based on the evaluation value, the feature quantity constraint prediction result, and the constraint violation warning value held in the driving command parameter determining unit 15B. The feature quantity predicting unit 14B determines the feature quantity prediction result and the constraint violation warning value based on the feature quantity model and the driving command parameter candidates.

[0160] The driving command parameter candidate generation unit 11B determines driving command parameter candidates based on the evaluation value and the constraint violation warning value. A plurality of pairs of command parameters and evaluation values ​​are held as initial points for determining the next driving command parameter candidates. The next driving command parameter candidates are determined based on the held plurality of pairs. The method of determining the next driving command parameter candidates may follow the golden search method.

[0161] The feature quantity prediction unit 14B calculates a predicted value of a feature quantity for the driving command parameter candidate based on the learned feature quantity model. The feature quantity prediction unit 14B outputs a feature quantity obtained by inputting the driving command parameter candidate to the learned feature quantity model as a feature quantity prediction result. The feature quantity prediction unit 14B also outputs the driving command parameter candidate. At this time, the feature quantity prediction unit 14B outputs the feature quantity prediction result and the driving command parameter candidate to each of the evaluation value determination unit 17B and the driving command parameter determination unit 15B.

[0162] The driving command parameter determination unit 15B can select a driving command parameter and generate a constraint violation warning value based on the driving command parameter candidate, the evaluation value determined by the evaluation value determination unit 17B, and the feature value determined by the feature value prediction unit 14B. The driving command parameter determination unit 15B can determine that the driving command parameter candidate satisfies the constraint condition when the feature value is equal to or greater than a predetermined threshold value, and can determine that the driving command parameter candidate does not satisfy the constraint condition when the feature value is less than the threshold value. The driving command parameter determination unit 15B can attach a label to the driving command parameter candidate as a result of the determination. For example, the label of the driving command parameter candidate that satisfies the constraint condition can be set to "1" and the label of the driving command parameter candidate that does not satisfy the constraint condition can be set to "0". The driving command parameter determination unit 15B can determine the worst evaluation value among the evaluation values ​​held by the evaluation value determination unit 17B as the constraint violation warning value.

[0163] 16 is a flowchart for explaining the operation of the parameter adjustment device 10B according to the third embodiment. First, the prior data acquisition unit 12 acquires prior data (step S300). The feature model generation unit 13 generates a feature model based on the prior data acquired by the prior data acquisition unit 12 (step S301), and outputs the generated feature model as a trained feature model.

[0164] The driving command parameter candidate generating unit 11B determines the next driving command parameter candidate (step S302). The feature quantity predicting unit 14B calculates a predicted value of the feature quantity of the next driving command parameter candidate based on the learned feature quantity model generated by the feature quantity model generating unit 13 (step S303), and outputs the calculated predicted value of the feature quantity to the driving command parameter determining unit 15B as a feature quantity prediction result.

[0165] The driving command parameter determination unit 15B judges whether or not the constraint condition is predicted to be satisfied based on the feature prediction result output by the feature prediction unit 14B (step S304). If the constraint condition is not predicted to be satisfied (step S304: No), the driving command parameter determination unit 15B assigns the evaluation value Q of the driving command parameter candidate to the worst value currently being searched (step S305), sets it as a constraint violation warning value for the driving command parameter candidate, and returns to the process of step S302.

[0166] If it is predicted that the constraint condition is satisfied (step S304: Yes), the driving command parameter determination unit 15B selects one from the driving command parameter candidates based on the feature amount prediction result, and sets the selected driving command parameter candidate as the driving command parameter to be used by the positioning control device 20 (step S306). The positioning control device 20 executes the positioning operation based on the set driving command parameter.

[0167] The sensor 50 acquires a sensor signal that is a time-series signal of the position of the movable part 40, and the parameter adjustment device 10B receives the sensor signal from the sensor 50 (step S307).

[0168] The evaluation value determination unit 17B determines an evaluation value based on the sensor signal acquired in the positioning operation (step S308). The evaluation value determination unit 17B holds the determined evaluation value.

[0169] The feature amount determining unit 18 determines the feature amount based on the sensor signal acquired in the positioning operation (step S309). The feature amount determining unit 18 holds the determined feature amount.

[0170] The operation command parameter determination unit 15B determines whether or not the end condition is satisfied (step S310). The end condition is the same as that in step S208 of FIG.

[0171] If the termination condition is not satisfied (step S310: No), the feature model generation unit 13 updates the feature model based on the feature held by the feature determination unit 18 (step S311), and returns to the process of step S302.

[0172] When the termination condition is satisfied (step S310: Yes), the operation command parameter determination unit 15B determines the adjustment result (step S312). For example, the operation command parameter determination unit 15B selects evaluation values ​​held in the evaluation value determination unit 17B whose corresponding feature values ​​exceed a threshold value, selects the best evaluation value from the selected evaluation values, and selects the operation command parameter candidate corresponding to the selected evaluation value as the operation command parameter to be used by the positioning control device 20, thereby determining the adjustment result.

[0173] Fig. 17 is a flow chart for explaining the details of step S302 in Fig. 16. Here, as an example, the next operation command parameter candidate is determined according to the golden section method. The golden section method is a kind of ternary search method that determines the next parameter to be evaluated based on four parameters stored internally and their respective evaluation values.

[0174] First, the driving command parameter candidate generating unit 11B determines whether four initial points have been set, that is, whether four pairs of command parameters and evaluation values ​​are held (step S320).

[0175] If four initial points are set (step S320: Yes), the driving command parameter candidate generator 11B arranges the four initial points in ascending order of command parameter value, and sets the arranged command parameters as points PcmdA, PcmdB, PcmdC, and PcmdD. Next, the driving command parameter candidate generator 11B compares the magnitudes of evaluation values ​​corresponding to the two central points PcmdB and PcmdC of the four sets, and selects the point PcmdE with the largest evaluation value from the two central points PcmdB and PcmdC (step S321). Next, the driving command parameter candidate generator 11B excludes the point farthest from the selected point PcmdE from the points PcmdA, PcmdB, PcmdC, and PcmdD (step S322).

[0176] The driving command parameter candidate generating unit 11B selects a new point PcmdF within the interval defined by the remaining three points (step S323) and adds it as the fourth pair held by the driving command parameter candidate generating unit 11B. The driving command parameter candidate generating unit 11B selects the point PcmdF such that the ratio of the lengths of the intervals defined by the four command parameters, including the command parameters included in the remaining three points and the command parameter of point PcmdF, is 1:γ:1. Here, γ is the reciprocal of the golden ratio.

[0177] The driving command parameter candidate generating unit 11B determines the command parameter corresponding to the point PcmdF as the next driving command parameter candidate (step S324), and ends the process.

[0178] If four initial points have not been set (step S320: No), the operation command parameter candidate generating unit 11B sets initial points for four parameters held internally, selects one of the four set initial points for which an evaluation value has not been calculated, and determines the parameter corresponding to the selected initial point as the next operation command parameter candidate (step S325).For example, the initial points are set to four points, including two points, the upper limit PcmdU and the lower limit PcmdL of the search range of the preset command parameters, and two points Pcmd1 and Pcmd2 that divide the interval between the upper limit PcmdU and the lower limit PcmdL at a ratio of 1:γ:1.

[0179] 17, for simplicity, the command parameters are considered as scalar values ​​and one-dimensional optimization using the golden search method is described, but the present disclosure can be implemented in the same way when the command parameters are vectors. As optimization methods in multidimensional parameter space, the coordinate descent method, Powell's method, Nelder-Mead method, particle swarm optimization method, etc. are known, and these methods may be applied.

[0180] In the third embodiment, when the driving command parameter candidate is not predicted to satisfy the constraint condition, the driving command parameter determination unit 15B determines the worst evaluation value among the evaluation values ​​held during the current search as the constraint violation warning value for the driving command parameter candidate. However, this embodiment is not limited to such an example. As a result, it is only necessary to adjust the driving command parameter candidate that does not satisfy the constraint condition so that the positioning control device 20 does not use it. For example, the constraint violation warning value may be a value indicating a worse value than the worst evaluation value among the evaluation values ​​held. Furthermore, the constraint violation warning value does not necessarily need to be generated in the form of an evaluation value. As another example, the constraint violation warning value may be generated as a signal instructing not to select the target driving command parameter candidate.

[0181] As described above, according to the third embodiment, the evaluation value determiner 17B acquires the feature amount prediction results in association with each of the driving command parameter candidates, and if the feature amount prediction result corresponding to the driving command parameter candidate satisfies a predetermined condition, maintains the evaluation value of the driving command parameter candidate, and if the feature amount prediction result corresponding to the driving command parameter candidate does not satisfy the condition, updates the evaluation value of the driving command parameter candidate based on the worst value of the evaluation values. As a result, information on the driving command parameter candidates predicted not to satisfy the constraint conditions is held, so that the number of times that driving command parameter candidates not satisfying the constraint conditions are generated is reduced, and it is possible to shorten the time required to determine the driving command parameters.

[0182] The configurations shown in the above embodiments are examples of the contents of the present disclosure, and may be combined with other known technologies, and parts of the configurations may be omitted or modified without departing from the gist of the present disclosure.

[0183] For example, the above operations described using the flowcharts are merely examples, and processing may be omitted, processing may be added, or the order in which processing is performed may be changed as appropriate.

[0184] In the above embodiment, control is mainly performed based on the position and movement distance of the movable part 40, but control may be performed based on the position and movement distance of the motor 30 instead of the movable part 40.

[0185] Various aspects of the present disclosure are summarized below as appendices.

[0186] (Appendix 1) A parameter adjustment device outputs an operation command parameter to a positioning control device having a command waveform generation unit that generates a position command, which is a time-series signal that commands a position of a motor that executes a positioning operation to move a movable part by a target moving distance, based on an operation command parameter, and a drive control unit that supplies power to drive the motor based on the position command, a driving command parameter candidate generating unit configured to generate a plurality of driving command parameter candidates that are candidates for the driving command parameters; a preliminary data acquisition unit that acquires preliminary data including a feature quantity that indicates a state of at least one of the positioning control device, the movable part, and the motor when the trial positioning operation is performed; a feature model generation unit that generates a trained feature model for inferring the feature from the driving command parameters by using the advance data; a feature prediction unit that inputs the plurality of driving command parameter candidates to the learned feature model and outputs, as a feature prediction result, a predicted value of the feature output for each of the plurality of driving command parameter candidates; a driving command parameter determination unit that selects one of the plurality of driving command parameter candidates based on the feature amount prediction result and determines the selected one as the driving command parameter to be used in generating the position command; A parameter adjustment device comprising: (Appendix 2) an evaluation value determination unit that determines an evaluation value of the positioning operation corresponding to the driving command parameters of the trial positioning operation based on a sensor signal that detects a position of the movable part or the motor when the trial positioning operation is performed using the driving command parameters, and determines the evaluation value corresponding to the driving command parameter candidates using a learned evaluation value model for inferring the evaluation value from the driving command parameter candidates; an evaluation value model generation unit that generates the trained evaluation value model by using the prior data; Equipped with The operation command parameter determination unit is the parameter adjusting device according to claim 1, further comprising: a plurality of operation command parameter candidates selected based on the feature amount prediction result and the evaluation value, and determining the selected candidate as the operation command parameter to be used by the positioning control device. (Appendix 3) the evaluation value determination unit acquires the feature quantity prediction results in association with the driving command parameter candidates, When the feature quantity prediction result corresponding to the driving command parameter candidate satisfies a predetermined condition, the value of the evaluation value of the driving command parameter candidate is maintained; and updating the evaluation value of the driving command parameter candidate based on a worst value of the evaluation values ​​when the feature amount prediction result corresponding to the driving command parameter candidate does not satisfy the condition. (Appendix 4) a feature determining unit that determines the feature for the operation command parameter corresponding to the trial positioning operation; Further equipped with 4. The parameter adjusting device according to claim 1, wherein the feature determining unit determines the feature based on a result of comparing the position command with a sensor signal that detects a position of the movable part or the motor. (Appendix 5) The evaluation value determination unit The parameter adjustment device according to claim 2 or 3, wherein the evaluation value is determined based on a settling time, which is a time required for a deviation between a detection value of a moving distance of the movable part and the target moving distance to become smaller than a predetermined allowable value. (Appendix 6) The advance data acquisition unit is at least one of the following is stored as the prior data in association with the feature quantity indicating the state of at least one of the positioning control device, the movable part, and a state of fixing the positioning device to a floor surface, the movable part, and the positioning device including a stand on which the movable part is movably installed, the ambient air temperature around the positioning device, the vibration of the floor surface on which the positioning device is installed, the usage time of the positioning device, and the state of a lubricant of the positioning device; Installing the positioning device at a first installation location and storing the acquired advance data as first advance data; 6. The parameter adjustment device according to claim 1, wherein the positioning device is installed at a second installation location different from the first installation location, and the advance data acquired is stored as second advance data. (Appendix 7) a motor that performs a positioning operation to move the movable part by a target moving distance; a positioning control device including a command waveform generating unit that generates a position command, which is a time-series signal that commands a position of the motor, based on an operation command parameter, and a drive control unit that supplies power to drive the motor based on the position command; a parameter adjustment device that outputs the operation command parameters to the positioning control device; Equipped with The parameter adjustment device includes: a driving command parameter candidate generating unit configured to generate a plurality of driving command parameter candidates that are candidates for the driving command parameters; a preliminary data acquisition unit that acquires preliminary data including a feature quantity that indicates a state of at least one of the positioning control device, the movable part, and the motor when the trial positioning operation is performed; a feature model generation unit that generates a trained feature model for inferring the feature from the driving command parameters by using the advance data; a feature prediction unit that inputs the plurality of driving command parameter candidates to the learned feature model and outputs, as a feature prediction result, a predicted value of the feature output for each of the plurality of driving command parameter candidates; a driving command parameter determination unit that selects one of the plurality of driving command parameter candidates based on the feature amount prediction result and determines the selected one as the driving command parameter to be used in generating the position command; A positioning system comprising: (Appendix 8) generating a plurality of operation command parameter candidates, which are candidates for operation command parameters to be output to a positioning control device, by the parameter adjustment device; acquiring, by the parameter adjustment device, advance data including feature quantities indicative of a state of at least one of the positioning control device, the movable part, and the motor when the motor performs a trial positioning operation in which the movable part is moved by a target moving distance; a step of the parameter adjustment device generating a learned feature model for inferring the feature from the driving command parameters by using the prior data; a step of the parameter adjustment device inputting a plurality of the driving command parameter candidates into the learned feature model and outputting, as a feature prediction result, a predicted value of the feature output for each of the plurality of driving command parameter candidates; a step of the parameter adjustment device selecting one of the plurality of driving command parameter candidates based on the feature amount prediction result and determining the selected one as the driving command parameter to be used by the positioning control device; a step of the positioning control device generating a position command which is a time-series signal for commanding a position of the motor based on the operation command parameters output by the parameter adjustment device; the positioning control device supplying power to drive the motor based on the position command; a step of executing the positioning operation by the motor to move the movable part by the target movement distance; A positioning method comprising: (Appendix 9) a gripping step in which the gripping portion grips the electronic component; a positioning step of moving the gripping portion holding the electronic component on a mounting board by a target movement distance; a placement step of placing the electronic component on the mounting board by the gripper releasing the electronic component at a position where the gripper stops; Including, The positioning step includes: generating a plurality of operation command parameter candidates, which are candidates for operation command parameters to be output to a positioning control device, by the parameter adjustment device; acquiring, by the parameter adjustment device, advance data including feature quantities indicative of a state of at least one of the positioning control device, the movable part, and the motor when the motor performs a trial positioning operation in which the movable part is moved by a target moving distance; a step of the parameter adjustment device generating a learned feature model for inferring the feature from the driving command parameters by using the prior data; a step of the parameter adjustment device inputting a plurality of the driving command parameter candidates into the learned feature model and outputting, as a feature prediction result, a predicted value of the feature output for each of the plurality of driving command parameter candidates; a step of the parameter adjustment device selecting one of the plurality of driving command parameter candidates based on the feature amount prediction result and determining the selected one as the driving command parameter to be used by the positioning control device; a step of the positioning control device generating a position command which is a time-series signal for commanding a position of the motor based on the operation command parameters output by the parameter adjustment device; the positioning control device supplying power to drive the motor based on the position command; a step of executing the positioning operation by the motor to move the movable part by the target movement distance; A method for manufacturing an electronic component mounting board, comprising: [Explanation of symbols]

[0187] 1, 1A, 1B Positioning system, 10, 10A, 10B Parameter adjustment device, 11, 11B Operation command parameter candidate generation unit, 12, 12A Advance data acquisition unit, 13 Feature quantity model generation unit, 14, 14A Feature quantity prediction unit, 15, 15A, 15B Operation command parameter determination unit, 16 Evaluation value model generation unit, 17, 17A, 17B Evaluation value determination unit, 18 Feature quantity determination unit, 20 Positioning control device, 21 Command waveform generation unit, 22 Drive control unit, 30 Motor, 40 Movable part, 50 Sensor, 90 Processing circuit, 91 Control circuit, 92 Processor, 93 Memory.

Claims

1. A parameter adjustment device that outputs the operation command parameters to a positioning control device having a command waveform generation unit that generates a position command, which is a time-series signal that commands the position of a motor that performs a positioning operation to move a movable part by a target travel distance, based on operation command parameters, and a drive control unit that supplies power based on the position command to drive the motor, A driver command parameter candidate generation unit generates a plurality of driver command parameter candidates, which are candidates for the aforementioned driver command parameters. A pre-data acquisition unit acquires pre-data including feature quantities that indicate the state of at least one of the positioning control device, the movable part, and the motor when the trial positioning operation is performed, A feature model generation unit generates a trained feature model for inferring the features from the driving command parameters using the aforementioned prior data, A feature prediction unit inputs a plurality of candidate driving command parameters into the trained feature model and outputs the predicted value of the feature output for each of the plurality of candidate driving command parameters as a feature prediction result. A driving command parameter determination unit selects one of the candidate driving command parameters based on the feature prediction results and determines it to be used as a driving command parameter for generating the position command. A parameter adjustment device characterized by comprising the following features.

2. The parameter adjustment device according to Claim 1, characterized in that the feature model generation unit generates a trained feature model, which is a statistical model for inferring the features from the driving command parameters, using the prior data.

3. An evaluation value determination unit that determines an evaluation value of the positioning operation corresponding to the operation command parameters of the trial positioning operation based on sensor signals detecting the position of the movable part or the motor when the trial positioning operation using the operation command parameters is performed, and determines the evaluation value corresponding to the candidate operation command parameters using a trained evaluation value model for inferring the evaluation value from the candidate operation command parameters, The feature model generation unit generates a trained feature model, which is a statistical model for inferring the predicted values ​​of the features and the confidence intervals of the predicted values. The feature prediction unit inputs a plurality of candidate driving command parameters into the trained feature model and outputs the predicted value of the feature output for each of the plurality of candidate driving command parameters, and at least one of the lower limit and upper limit of the confidence interval as the feature prediction result. The parameter adjustment device according to claim 2, characterized in that the operation command parameter determination unit selects one of the multiple operation command parameter candidates based on the feature prediction result and the evaluation value, and determines it as the operation command parameter to be used by the positioning control device.

4. An evaluation value determination unit determines an evaluation value of the positioning operation corresponding to the operation command parameters of the trial positioning operation based on sensor signals detecting the position of the movable part or the motor when the trial positioning operation using the operation command parameters is performed, and determines the evaluation value corresponding to the candidate operation command parameters using a trained evaluation value model for inferring the evaluation value from the candidate operation command parameters. An evaluation value model generation unit generates the pre-trained evaluation value model using the aforementioned prior data, Equipped with, The aforementioned operation command parameter determination unit is: The parameter adjustment device according to claim 3, characterized in that, based on the feature prediction results and the evaluation values, one is selected from a plurality of candidate operating command parameters and determined to be the operating command parameter used by the positioning control device.

5. The evaluation value determination unit acquires the feature prediction results and associates them with each of the candidate driving command parameters. If the feature prediction result corresponding to the candidate driving command parameter satisfies the conditions, the value of the evaluation value of the candidate driving command parameter is maintained. The parameter adjustment device according to claim 4, characterized in that, if the feature prediction result corresponding to the candidate driving command parameter does not satisfy the conditions, the evaluation value of the candidate driving command parameter is updated based on the worst value among the evaluation values.

6. A feature determination unit that determines the feature quantities for the operation command parameters corresponding to the trial positioning operation, Furthermore, The parameter adjustment device according to claim 1 or 3, characterized in that the feature determination unit determines the feature based on the result of comparing the position command with the sensor signal detecting the position of the movable part or the motor.

7. The aforementioned evaluation value determination unit, The parameter adjustment device according to claim 3 or 4, characterized in that the evaluation value is determined based on a settling time, which is the time until the magnitude of the deviation between the detected value of the travel distance of the movable part and the target travel distance becomes smaller than a predetermined allowable value.

8. The aforementioned pre-data acquisition unit, The positioning device comprising the motor, the movable part, and the frame on which the movable part is movably mounted is fixed to the floor, the ambient temperature around the positioning device, the vibration of the floor on which the positioning device is installed, the operating time of the positioning device, and the state of the lubricant in the positioning device are stored as prior data in association with the feature quantity indicating the state of at least one of the positioning control device, the movable part, and the motor. The positioning device is installed at the first installation location, and the acquired prior data is retained as the first prior data. The parameter adjustment device according to claim 1 or 2, characterized in that the positioning device is installed at a second installation location different from the first installation location, and the prior data obtained is retained as second prior data.

9. A motor that performs a positioning operation to move the movable part by a target distance, A positioning control device having a command waveform generation unit that generates a position command, which is a time-series signal that commands the position of the motor, based on operating command parameters, and a drive control unit that drives the motor by supplying power based on the position command, A parameter adjustment device that outputs the operation command parameters to the positioning control device, Equipped with, The parameter adjustment device is A driver command parameter candidate generation unit generates a plurality of driver command parameter candidates, which are candidates for the aforementioned driver command parameters. A pre-data acquisition unit acquires pre-data including feature quantities that indicate the state of at least one of the positioning control device, the movable part, and the motor when the trial positioning operation is performed, A feature model generation unit generates a trained feature model for inferring the features from the driving command parameters using the aforementioned prior data, A feature prediction unit inputs a plurality of candidate driving command parameters into the trained feature model and outputs the predicted value of the feature output for each of the plurality of candidate driving command parameters as a feature prediction result. A driving command parameter determination unit selects one of the candidate driving command parameters based on the feature prediction results and determines it to be used as a driving command parameter for generating the position command. A positioning system characterized by having the following features.

10. The parameter adjustment device generates multiple candidate driving command parameters, which are candidates for driving command parameters to be output to the positioning control device. The parameter adjustment device acquires prior data including feature quantities that indicate the state of at least one of the positioning control device, the movable part, and the motor when the motor experimentally performs a positioning operation in which the motor moves the movable part by a target travel distance. The parameter adjustment device includes the step of generating a trained feature model for inferring the features from the driving command parameters using the prior data, The parameter adjustment device inputs a plurality of candidate driving command parameters into the trained feature model and outputs the predicted value of the feature output for each of the plurality of candidate driving command parameters as a feature prediction result. The parameter adjustment device selects one of the multiple candidate operating command parameters based on the feature prediction result and determines it to be the operating command parameter used by the positioning control device. The positioning control device generates a position command, which is a time-series signal that commands the position of the motor, based on the operation command parameters output by the parameter adjustment device. The positioning control device drives the motor by supplying power based on the position command, The motor performs the positioning operation, moving the movable part by the target travel distance, A positioning method characterized by including

11. A gripping process in which the gripping part grips the electronic component, A positioning step of moving the gripping portion, which is gripping the electronic component on the mounting substrate, by a target distance, The arrangement step involves releasing the electronic component from the gripping portion at the position where the gripping portion has stopped, thereby placing the electronic component on the mounting substrate. Includes, The positioning step is, The parameter adjustment device generates multiple candidate driving command parameters, which are candidates for driving command parameters to be output to the positioning control device. The parameter adjustment device acquires prior data including feature quantities that indicate the state of at least one of the positioning control device, the movable part, and the motor when the motor experimentally performs a positioning operation in which the motor moves the movable part by a target travel distance. The parameter adjustment device includes the step of generating a trained feature model, which is a statistical model for inferring the features from the driving command parameters, using the prior data. The parameter adjustment device inputs a plurality of candidate driving command parameters into the trained feature model and outputs the predicted value of the feature output for each of the plurality of candidate driving command parameters as a feature prediction result. The parameter adjustment device selects one of the multiple candidate operating command parameters based on the feature prediction result and determines it to be the operating command parameter used by the positioning control device. The positioning control device generates a position command, which is a time-series signal that commands the position of the motor, based on the operation command parameters output by the parameter adjustment device. The positioning control device drives the motor by supplying power based on the position command, The motor performs the positioning operation, moving the movable part by the target travel distance, A method for manufacturing an electronic component mounting substrate, characterized by including the following:

Images