Control system, learning device, inference device, and controller

Abstract

This control system comprises a detection unit that detects a position of a workpiece moving on a conveyance path, a drive unit that drives a processing unit for processing the workpiece, a controller (12) that controls the drive unit, and a remote I / O that outputs a detection signal from the detection unit to the controller (12), wherein: the controller (12) is provided with a Kalman filter (23) for estimating a future position of the workpiece on the basis of the detection signal, and a timing generation unit (25) for generating a processing timing by the processing unit on the basis of the estimated future position of the workpiece; and the controller (12) transmits the processing timing to the drive part via the remote I / O.

Description

Control system, learning device, inference device and controller

[0001] The present disclosure relates to a control system, a learning device, a reasoning device, and a controller.

[0002] In the cutting process of workpieces in equipment consisting of an unwinding shaft and a winding shaft, such as packaging machines and printing machines, an optical sensor reads a mark called a registration mark, which is pre-printed on the workpiece as it moves at high speed, and then the workpiece is cut by a device equipped with a motor after passing a certain position. Here, since the reading position and the cutting position are far apart, it is necessary to detect the registration mark and estimate the future position based on the state of the equipment at the time the reading was performed.

[0003] Patent Document 1 discloses that a mark detector is disposed upstream of a rotating member that cuts the transported web material, the detection signal of the mark detector is sent to a calculation control unit, the feed length is calculated from the speed command and the mark detection signal, and the phase of the timing of the cutting process is adjusted.

[0004] Japanese Patent Application Publication No. 11-236153

[0005] However, conventional methods estimate future positions by linear extrapolation, which results in poor estimation accuracy in acceleration / deceleration regions other than constant velocity regions, resulting in deviations in the cutting position. Furthermore, since filters cannot be freely switched, it is not possible to select the optimal extrapolation method depending on different conditions, which results in the inability to improve product quality.

[0006] The present disclosure has been made in consideration of the above circumstances, and aims to provide a future position estimation accuracy that exceeds linear extrapolation in the acceleration / deceleration region.

[0007] In order to achieve the above object, the control system of the present disclosure comprises a detection unit that detects the position of a workpiece moving on a conveying path, a drive unit that drives a processing unit that processes the workpiece, a controller that controls the drive unit, and a remote I / O that outputs a detection signal from the detection unit to the controller, and the controller comprises a Kalman filter that estimates the future position of the workpiece based on the detection signal, and a timing generation unit that generates processing timing for the processing unit based on the estimated future position of the workpiece, and transmits the processing timing to the drive unit via the remote I / O.

[0008] According to the present disclosure, by using a Kalman filter to estimate the future position of a workpiece, it is possible to achieve a higher accuracy in estimating the future position than linear extrapolation in the acceleration / deceleration region.

[0009] FIG. 1 is a diagram showing the configuration of a control device according to a first embodiment of the present disclosure. FIG. 2 is a diagram showing the functional configuration of a controller of the control device according to the first embodiment of the present disclosure. FIG. 3 is a diagram showing the configuration of a Kalman filter. FIG. 4 is a waveform diagram of an estimated value calculated using a Kalman filter and a linear extrapolator during an acceleration period. FIG. 5 is a diagram showing the hardware configuration of a controller of the control device according to the first embodiment of the present disclosure.

[0010] First Embodiment A control system 100 according to a first embodiment of the present disclosure will now be described with reference to the drawings. In the drawings, the same or equivalent parts are denoted by the same reference numerals.

[0011] 1 is a configuration diagram showing a control system 100 according to a first embodiment. The control system 100 includes an unwinding shaft 1, a winding shaft 2, a workpiece 3, which is a sheet-like material such as packaging paper, photographic paper, or film transported between the unwinding shaft 1 and the winding shaft 2, a cutter 4 for cutting the workpiece 3, a cylindrical rotor 5 having an outer circumferential surface on which the cutter 4 is provided and which rotates in the transport direction, a servo motor 6 for rotating the rotor 5, a servo amplifier 7 for driving the servo motor 6, a mark sensor 9 for reading a mark 8 provided on the surface of the workpiece 3 and generating a calculation start signal for the cutting timing, an encoder 10 for detecting the movement of the workpiece 3, a remote I / O 11 for receiving detection signals from the mark sensor 9 and the encoder 10, and a controller 12 for estimating the cutting position of the workpiece 3 and controlling the cutting. The controller 12, the remote I / O 11, and the servo amplifier 7 are connected by a communication line 13. The cutter 4 is an example of a processing unit of the present disclosure, the servo motor 6 is an example of a drive unit of the present disclosure, and the mark sensor 9 and the encoder 10 are an example of a detection unit of the present disclosure.

[0012] A workpiece 3 is wound in a roll around the unwinding shaft 1, and is unwound by the rotation of the winding shaft 2. The winding shaft 2 winds up the workpiece 3 unwound from the unwinding shaft 1. The workpiece 3 unwound from the unwinding shaft 1 is transported toward a rotor 5 on which a cutter 4 is provided by the rotation of the unwinding shaft 1 and the winding shaft 2.

[0013] Black marks 8 are provided at regular intervals along the edge of the surface of the workpiece 3 in the conveyance direction. The marks 8 indicate the cutting positions of the workpiece 3, and dotted lines indicate the cutting lines when the cutter 4 cuts from the marks 8. Note that the cutting lines are shown in the drawings for the purpose of explanation, and are not actually provided on the surface of the workpiece 3.

[0014] The cutter 4 cuts the workpiece 3 and is provided on the outer circumferential surface of the rotor 5 in parallel to the rotation axis. The rotor 5 rotates in the conveying direction of the workpiece 3 at a position where its outer circumferential surface faces the workpiece 3 due to the rotation of the servo motor 6. When the cutter 4 reaches a position facing the workpiece 3 due to the rotation of the rotor 5, it comes into contact with the workpiece 3 and cuts it.

[0015] The servo motor 6 is driven by a servo amplifier 7, which controls the servo motor 6 in accordance with the stored operating parameters. Note that, although the present embodiment discloses a configuration in which the rotor 5 with the cutter 4 attached is rotated by a rotary motor to cut the workpiece 3, the present invention is not limited to this, and for example, a configuration in which the cutter 4 is moved parallel to the conveying direction of the workpiece 3 by a linear motor may also be used.

[0016] The remote I / O 11 receives signals input from the mark sensor 9 and the encoder 10. The mark sensor 9 is an optical sensor that is disposed above the conveyance path of the workpiece 3, upstream of the rotor 5, and detects marks 8 provided on the surface of the workpiece 3. The encoder 10 is disposed on the conveyance path of the workpiece 3 and outputs pulses continuously in accordance with the movement of the workpiece 3. The remote I / O 11 communicates with the controller 12 and operates in accordance with instructions sent from the controller 12. The remote I / O 11 transmits signals input from the mark sensor 9 and the encoder 10 to the controller 12. The remote I / O 11 also outputs signals to the servo amplifier 7 in accordance with instructions from the controller 12.

[0017] The controller 12 is, for example, a PLC (Programmable Logic Controller). The controller 12 acquires the current position of the workpiece 3 by receiving detection signals from the mark sensor 9 and the encoder 10. Based on the acquired current position, the controller 12 obtains a predicted future position of the workpiece 3, predicts the position at which the workpiece 3 will be cut, and outputs a signal instructing the timing of cutting the workpiece 3 to the remote I / O 11.

[0018] 2 shows a functional configuration representing the internal processing of the controller 12. The controller 12 includes a pulse counter 21 that counts encoder pulses, a switch 22 that switches the direction in which the count value of the pulse counter 21 is output, a Kalman filter 23 and a linear extrapolator 24 that are estimators that predict the future position of the workpiece 3, and a timing generation unit 25 that generates a cut timing signal for cutting the workpiece 3.

[0019] The pulse counter 21 receives the mark detection signal, which is the output of the mark sensor 9, and the encoder pulse, which is the output of the encoder 10, from the remote I / O 11. The pulse counter 21 starts counting the encoder pulses upon receiving the mark detection signal, acquires a current position signal indicating the current position of the workpiece 3, and outputs the signal to the switch 22.

[0020] The switch 22 selects an output destination of the input current position signal and outputs it. The output destinations of the switch 22 are connected to a Kalman filter 23 and a linear extrapolator 24. The switch 22 switches the output destination between the Kalman filter 23 and the linear extrapolator 24 in accordance with a switching signal. The output timing of the switching signal is set in advance by the user.

[0021] The Kalman filter 23 and linear extrapolator 24 estimate and calculate the future position of the workpiece 3 based on the current position signal input from the switch 22. The linear extrapolator 24 estimates the future position by connecting the current position and the past position with a straight line and linearly extrapolating using that line. Linear extrapolation has the advantage of being simple and fast to calculate and requiring few resources. However, in the case of a data pattern that changes in a curved manner, it is not possible to accurately estimate the future position. In contrast, the Kalman filter 23 can accurately estimate the future position even for a data pattern that changes in a curved manner.

[0022] FIG. 3 shows the details of the Kalman filter 23. The Kalman filter 23 includes a delayer 31, a predictor 32, and a smoother 33. The delayer 31 holds the state estimate from the previous sampling for one sampling period and outputs it for the current sampling period. The state estimate includes a smoothed value and a smoothed error covariance matrix. The predictor 32 calculates a predicted value and a prediction error covariance matrix using the smoothed value and the smoothed error covariance matrix from the previous sampling period obtained by the delayer 31. The calculated predicted value and prediction error covariance matrix are output to the smoother 33. The smoother 33 acquires the mark detection signal and the current position signal calculated from the encoder pulse as observation data, and calculates the filter gain, smoothed value, and smoothed error covariance matrix. The calculation of the filter gain, smoothed value, and smoothed error covariance matrix is ​​provided by the predictor 32 in addition to the mark detection signal and the current position signal calculated from the encoder pulse, and the processing results of the predictor 32 are also provided. The processing results of the smoother 33 are then returned to the delayer 31. The future position predictor 34 calculates the future position after N samplings by extrapolation from the smoothed position and smoothed velocity vector obtained by the smoother 33 .

[0023] The workpiece 3 moving along the conveying path moves at a speed according to a predetermined speed map. Here, the speed map may include, for example, an acceleration period immediately after the start of movement, in which the workpiece 3 accelerates until it reaches a predetermined speed, a constant-speed period in which the workpiece 3 maintains that speed once it reaches the predetermined speed, and a deceleration period in which the workpiece 3 decelerates. FIG. 4 shows waveforms of estimated values ​​calculated using the Kalman filter 23 and the linear extrapolator 24 during the acceleration period. In FIG. 4, "a" indicates an estimated value calculated using a linear Kalman filter as the Kalman filter 23, and "b" indicates an estimated value calculated using the linear extrapolator 24. As a result, in regions involving acceleration and deceleration, the estimated value calculated using the Kalman filter 23 produces smoother estimation results than the estimated value calculated using the linear extrapolator 24. Therefore, during acceleration and deceleration periods in which the position of the workpiece 3 changes curvilinearly, the future position can be estimated accurately by using the Kalman filter 23. On the other hand, during constant-speed periods in which the position of the workpiece 3 changes linearly, the linear extrapolator 24 can also produce highly accurate estimations. Therefore, by using the linear extrapolator 24 during the constant velocity period, it is possible to accurately estimate the future position while taking into consideration the calculation speed, resource load, etc. Therefore, when setting the output timing of the switching signal, the user sets the Kalman filter 23 to be selected during the acceleration period when the workpiece 3 accelerates and the deceleration period when the workpiece 3 decelerates, and sets the linear extrapolator 24 to be selected during the constant velocity period when the workpiece 3 moves at a constant velocity. The acceleration / deceleration period and the constant velocity period may be determined by referring to a velocity map, or may be determined by determining the velocity, acceleration, or jerk from the encoder pulse.

[0024] The timing generation unit 25 generates a target timing for cutting the workpiece 3 based on the predicted future position information estimated by the Kalman filter 23 or the linear extrapolator 24. Specifically, the timing generation unit 25 calculates the time it takes for the workpiece 3 to reach the cutting position by dividing the estimated predicted future position by the transport speed of the workpiece 3, and further calculates how many clocks later the arrival time will be. The generated cut timing signal is output to the servo amplifier 7 via the remote I / O 11. The servo amplifier 7 controls the rotation of the rotor 5 based on the cut timing signal, and cuts the workpiece 3 with the cutter 4.

[0025] 5, the controller 12 includes a processor 41 that processes data according to a control program, a main memory 42 that functions as a work area for the processor, an auxiliary memory 43 for long-term data storage, an input unit 44 that accepts data input, an output unit 45 that outputs data, a communication unit 46 that communicates with other devices, a display unit 47, and a bus that interconnects these elements. The auxiliary memory 43 stores a control program that controls the timing of cutting executed by the processor. The input unit 44 receives signals from the mark sensor 9 and the encoder 10 sent from the remote I / O 11 and provides the signals to the processor 41.

[0026] (Second embodiment) In order to estimate the future position with higher accuracy, the set values ​​of parameters such as the machine model and Kalman gain used in the Kalman filter 23 must match the physical quantities of the actual machine. In general, identifying these parameters for a complex machine is difficult and may be determined experimentally. In this embodiment, optimal parameters for the Kalman filter 23 are determined by learning using the position, velocity, and acceleration of the workpiece 3 as inputs, with cutting accuracy as a reward. Learning is performed using machine learning based on reinforcement learning (Q-learning).

[0027] 6 is a configuration diagram of a machine learning device for the control system 100. The learning device 201 includes a data acquisition unit 202 and a model generation unit 203.

[0028] The data acquisition unit 202 acquires the parameters of the Kalman filter 23 as the action A, and the position, velocity, and acceleration of the workpiece 3 as the state S as learning data.

[0029] The model generation unit 203 learns the parameters of the Kalman filter 23 for obtaining the most accurate cutting results as the optimal action A, based on learning data including the parameters of the Kalman filter 23 as the action A and the position, speed, and acceleration of the workpiece 3 as the state S. That is, it generates a learned model that infers the parameters of the Kalman filter 23 as the optimal action A from the position, speed, and acceleration of the workpiece 3, which is the state S of the control system 100.

[0030] The learning algorithm used by the model generation unit 203 may be a known algorithm such as supervised learning, unsupervised learning, or reinforcement learning. As an example, a case where reinforcement learning is applied will be described. In reinforcement learning, an agent (acting subject) in a certain environment observes the current state (environmental parameters) and determines the action to be taken. The environment changes dynamically depending on the agent's actions, and the agent is given a reward according to the change in the environment. The agent repeats this process and learns the course of action that will obtain the most reward through a series of actions. Q-learning and TD-learning are known as representative reinforcement learning methods. For example, in the case of Q-learning, a general update formula for the action value function Q(s, a) is expressed as Equation (1).

[0031]

[0032] In formula (1), s ｔ represents the state of the environment at time t, and a ｔ represents the action at time t. ｔ Therefore, the state is s ｔ＋１ It changes to r ｔ＋１ represents the reward that can be obtained depending on the change in state, γ represents the discount rate, and α represents the learning coefficient. Note that γ is in the range of 0<γ≦1, and α is in the range of 0<α≦1. ｔ and state S becomes state sｔ and the state s at time t ｔ Best action in a ｔ Learn.

[0033] The update formula expressed by equation (1) increases the action value Q if the action value Q of the action a with the highest Q value at time t+1 is greater than the action value Q of the action a executed at time t, and decreases the action value Q in the opposite case. In other words, the action value function Q(s, a) is updated so that the action value Q of the action a at time t approaches the best action value at time t+1. As a result, the best action value in a certain environment is propagated sequentially to the action values ​​in previous environments.

[0034] As described above, when generating a trained model by reinforcement learning, the model generation unit 203 includes a reward calculation unit 204 and a function update unit 205.

[0035] The reward calculation unit 204 calculates a reward based on the parameters of the Kalman filter 23, which is action A, and the position, speed, and acceleration of the workpiece 3, which is state S. The reward calculation unit 204 calculates a reward r based on the cutting result of the workpiece 3 as a reward criterion, i.e., the error range of how far the position where the workpiece 3 is cut is from a predetermined optimal cutting position. Therefore, the reward criterion is whether the cutting result of the workpiece 3 is within the error range of the standard. If the cutting result of the workpiece 3 is within the error range of the standard (reward increase criterion), the reward r is increased (for example, a reward of "1"), while if the cutting result of the workpiece 3 is beyond the error range of the standard (reward decrease criterion), the reward r is decreased (for example, a reward of "-1" is given).

[0036] The function update unit 205 updates the function for determining the optimal action A according to the reward calculated by the reward calculation unit 204, and outputs the updated function to the learned model storage unit 206. For example, in the case of Q-learning, the action value function Q(s ｔ , a ｔ ) is used as a function to calculate the optimal action A.

[0037] The learning process is repeated as described above. The learned model storage unit 206 stores the action value function Q(s) updated by the function update unit 205. ｔ , a ｔ ), i.e., stores the trained model.

[0038] Next, the learning process of the learning device 201 will be described with reference to Fig. 7. Fig. 7 is a flowchart showing the learning process of the learning device 201.

[0039] In step S11, the data acquisition unit 202 acquires the parameters of the Kalman filter 23, which is action A, and the position, velocity, and acceleration of the workpiece 3, which is state S, as learning data.

[0040] In step S12, the model generation unit 203 calculates a reward based on the action A and the state S. Specifically, the reward calculation unit 204 acquires the action A and the state S, and determines whether to increase or decrease the reward based on a predetermined reward standard.

[0041] If the remuneration calculation unit 204 determines that the remuneration should be increased (step S12: Yes), it increases the remuneration in step S13. On the other hand, if the remuneration calculation unit 204 determines that the remuneration should be decreased (step S12: No), it decreases the remuneration in step S14.

[0042] In step S15, the function update unit 205 updates the action value function Q(s) represented by equation (1) stored in the trained model storage unit 206 based on the reward calculated by the reward calculation unit 204. ｔ , a ｔ ) to update.

[0043] The learning device 201 repeatedly executes the above steps S11 to S15 to generate the action value function Q(s ｔ , a ｔ ) is stored as a trained model.

[0044] The learning device 201 in this embodiment is configured to store the learned model in a learned model storage unit 206 provided outside the learning device 201, but the learned model storage unit 206 may also be provided inside the learning device 201.

[0045] 8 is a configuration diagram of an inference device 301 related to the control system 100. The inference device 301 includes a data acquisition unit 302 and an inference unit 303.

[0046] The data acquisition unit 302 acquires the position, velocity, and acceleration of the workpiece 3 in state S.

[0047] The inference unit 303 uses the trained model to infer the parameters of the Kalman filter 23, which are the optimal action A. That is, by inputting the state S acquired by the data acquisition unit 302 into this trained model, it is possible to infer the optimal action A suited to the state S. The state S input here is data including the current position, velocity, and acceleration of the workpiece 3.

[0048] In this embodiment, the optimal action A is output using a trained model trained by the model generation unit 203 of the learning device 201 related to the control system 100, but it is also possible to obtain a trained model from another control system 100 and output the optimal action A based on this trained model.

[0049] Next, with reference to FIG. 9, a process for obtaining the optimal action A, that is, the optimal parameters of the Kalman filter 23, using the learning device 201 and determining the optimal cutting position will be described.

[0050] In step S21, the data acquisition unit 302 acquires the state S.

[0051] In step S22, the inference unit 303 inputs the state S into the learned model stored in the learned model storage unit 206, and obtains an optimal action A. The inference unit 303 outputs the obtained optimal action A to the controller 12 of the control system 100 (step S23).

[0052] In step S24, the controller 12 determines the optimum cutting position of the workpiece 3 using the output optimum action A, that is, the parameters of the Kalman filter 23 that can obtain a highly accurate cutting result.

[0053] In this embodiment, a case where reinforcement learning is applied to the learning algorithm used by the inference unit 303 has been described, but the present invention is not limited to this. As for the learning algorithm, other than reinforcement learning, supervised learning, unsupervised learning, semi-supervised learning, or the like can also be applied.

[0054] Furthermore, the learning algorithm used in the model generation unit 203 may be deep learning, which learns to extract the parameters of the Kalman filter 23, or machine learning may be performed according to other known methods, such as neural networks, genetic programming, functional logic programming, and support vector machines.

[0055] The learning device 201 and the inference device 301 may be connected to the controller 12 of the control system 100 via a network, for example, and may be devices separate from the control system 100. The learning device 201 and the inference device 301 may also be built into the controller 12. Furthermore, the learning device 201 and the inference device 301 may exist on a cloud server.

[0056] Furthermore, the model generation unit 203 may learn the optimal behavior A using learning data acquired from multiple control systems 100. Note that the model generation unit 203 may acquire learning data from multiple control systems 100 used in the same area, or may learn the optimal behavior A using learning data collected from multiple control systems 100 operating independently in different areas. It is also possible to add or remove a control system 100 that collects learning data to or from the target during the process. Furthermore, the learning device 201 that has learned the optimal behavior A for a certain control system 100 may be applied to another control system 100, and the optimal behavior A for the other control system 100 may be re-learned and updated.

[0057] In the above embodiment, since the system model is linear, a linear Kalman filter is used as the Kalman filter 23. On the other hand, when a nonlinear model is used, accurate position estimation is possible by using a nonlinear Kalman filter as the Kalman filter 23. Furthermore, when it is unclear whether the model is linear or nonlinear, position estimation compatible with both models is possible by using an extended Kalman filter as the Kalman filter 23. Furthermore, a plurality of these Kalman filters may be provided as the Kalman filter 23, and switching may be performed by the switch 22. Furthermore, this switching may be determined by obtaining the velocity, acceleration, or jerk from the encoder pulses.

[0058] In the above embodiment, cutting is described as the processing of the workpiece 3, but the present invention is not limited to this and may be applied to, for example, printing on the workpiece 3.

[0059] The present disclosure allows various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Furthermore, the above-described embodiments are intended to explain the present disclosure and do not limit the scope of the present disclosure. In other words, the scope of the present disclosure is defined by the claims, not the embodiments. Various modifications made within the scope of the claims and the meaning of equivalent disclosures are considered to be within the scope of the present disclosure.

[0060] The present disclosure can be widely applied to a control system that controls a processing position for processing a workpiece based on the position of the workpiece moving on a transport path.

[0061] 1 Unwinding shaft, 2 Winding shaft, 3 Workpiece, 4 Cutter, 5 Rotor, 6 Servo motor, 7 Servo amplifier, 8 Mark, 9 Mark sensor, 10 Encoder, 11 Remote I / O, 12 Controller, 13 Communication line, 21 Pulse counter, 22 Switching device, 23 Kalman filter, 24 Linear extrapolator, 25 Timing generation unit, 31 Delay unit, 32 Predictor, 33 Smoothing device, 34 Future position predictor, 41 Processor, 42 Main memory unit, 43 Auxiliary memory unit, 44 Input unit, 45 Output unit, 46 Communication unit, 47 Display unit, 100 Control system, 201 Learning device, 202 Data acquisition unit, 203 Model generation unit, 204 Reward calculation unit, 205 Function update unit, 206 Learned model storage unit, 301 Inference device, 302 Data acquisition unit, 303 Reasoning part.

Claims

1. A control system comprising: a detection unit that detects the position of a workpiece moving on a conveying path; a drive unit that drives a processing unit that processes the workpiece; a controller that controls the drive unit; and a remote I / O that outputs detection signals from the detection unit to the controller, wherein the controller comprises: a Kalman filter that estimates the future position of the workpiece based on the detection signals; and a timing generation unit that generates processing timing for the processing unit based on the estimated future position of the workpiece, and transmits the processing timing to the drive unit via the remote I / O.

2. The control system according to claim 1, wherein the controller comprises: a linear extrapolator that estimates the future position of the workpiece by linear extrapolation; and a switch that selects either the Kalman filter or the linear extrapolator and outputs the estimated future position of the workpiece to the timing generation unit.

3. The control system according to claim 2, wherein the controller instructs the switch to select the Kalman filter when the movement of the workpiece is accelerating or decelerating, and to select the linear extrapolator when the movement of the workpiece is at a constant speed.

4. The control system according to claim 3, wherein the controller determines the movement state of the workpiece from the speed, acceleration or jerk of the workpiece and instructs the switch to make a selection.

5. A control system according to any one of claims 1 to 4, wherein the workpiece is a sheet-like material, and the processing unit is a cutter that cuts the sheet-like material.

6. A control system according to any one of claims 1 to 5, wherein the detection unit is a mark detection sensor that generates a mark detection signal by detecting a mark provided on the workpiece.

7. A learning device comprising: a data acquisition unit for acquiring learning data including the position, speed, and acceleration of a workpiece and parameters of a Kalman filter in the control system of claim 1; and a model generation unit for using the learning data to generate a trained model for inferring parameters of the Kalman filter for obtaining the most accurate cutting results from the position, speed, and acceleration of the workpiece.

8. An inference device comprising: a data acquisition unit for acquiring the current position, speed, and acceleration of a workpiece in the control system of claim 1; and an inference unit for outputting parameters of a Kalman filter from the current position, speed, and acceleration of the workpiece using a learning model for inferring parameters of the Kalman filter for obtaining the most accurate cutting results from the position, speed, and acceleration of the workpiece.

9. A controller that receives a detection signal from a detection unit that detects the position of a workpiece moving on a conveying path via a remote I / O and controls a drive unit that drives a processing unit that processes the workpiece, the controller comprising: a Kalman filter that estimates the future position of the workpiece based on the detection signal; and a timing generation unit that generates processing timing for the processing unit based on the estimated future position of the workpiece, and transmits the processing timing to the drive unit via the remote I / O.

Images