JP2025099770A - Sheet feeding device, method and program, and learning model - Google Patents

Abstract

To provide a sheet feeding device, a method, and a program and a learning model for preventing idle feeding and double feeding by performing more optimal adjustment of a control parameter.SOLUTION: A sheet feeding device comprises: a loading base 10 for loading a plurality of sheets; a main floating air supply part 71 for floating the sheet by supplying air toward an end surface of the plurality of sheets loaded on the loading base 10; a suction conveying part 20 for sucking and conveying an uppermost sheet out of the plurality of sheets floated by air supply; a photographing part 40 for photographing a floating state of the plurality of sheets by air supply; a control parameter adjustment part 70 for adjusting a control parameter of the main floating air supply part 71 on the basis of information of the floating state acquired from a photographed image photographed by the photographing part 40; and a control part 61 for controlling the main floating air supply part 71 on the basis of the control parameter adjusted by the control parameter adjustment part 70.SELECTED DRAWING: Figure 2

Description

The present invention relates to a sheet supplying device, method, program, and learning model that floats sheets by supplying air and then adsorbs and transports the topmost sheet to supply it.

A conventional paper feeder that supplies paper to a printing device or the like is one that supplies air to multiple sheets of paper stacked on a stacking tray to raise them, and then sucks and transports the topmost sheet of the sheets to supply them.

For example, Patent Document 1 proposes a method for determining control parameters for a paper feeder such as the one described above based on paper physical property information.

JP 2023-16933 A

However, in Patent Document 1, although paper physical property information is used to adjust the control parameters, the actual floating state of the paper is not taken into consideration, so the optimal control parameters may not necessarily be set, which may result in empty feeding or double feeding.

In view of the above circumstances, the present invention aims to provide a sheet feeding device, method, program, and learning model that can adjust control parameters more optimally and prevent empty feeding and double feeding.

The sheet supplying device of the present invention includes a loading platform on which multiple sheets are loaded, a floating air supplying unit that floats the multiple sheets loaded on the loading platform by supplying air toward the end faces of the multiple sheets, a suction and transporting unit that adsorbs and transports the uppermost sheet of the multiple sheets floated by the air supply, an image capturing unit that captures an image of the floating state of the multiple sheets due to the air supply, a control parameter adjusting unit that adjusts the control parameters of the floating air supplying unit based on information on the floating state obtained from the image captured by the image capturing unit, and a control unit that controls the floating air supplying unit based on the control parameters adjusted by the control parameter adjusting unit.

The sheet supply device of the present invention photographs the floating state of multiple sheets due to the supply of air, and adjusts the control parameters of the floating air supply unit based on the information on the floating state of the sheets obtained from the photographed image. This makes it possible to adjust the control parameters more optimally, taking into account the actual floating state of the sheets, and prevents empty feeding and double feeding.

FIG. 1 is a block diagram showing a schematic configuration of an air sheet feeding system using an embodiment of a sheet feeding device of the present invention; FIG. 2 is a diagram showing a specific configuration of a paper feed device in the air paper feed system shown in FIG. 1 . FIG. 2 is a diagram showing a specific configuration of a paper feed device in the air paper feed system shown in FIG. 1 . Block diagram showing the control system of the paper feeder A diagram showing the shutter configuration A diagram showing an example of environmental parameters for reinforcement learning A diagram showing an example of the reward plus condition of the environmental parameters FIG. 13 is a diagram showing an example of adjustment amounts of control parameters; A flowchart for explaining a method for adjusting control parameters of a paper feeder using a learning model. FIG. 2 illustrates an embodiment of the air feeding system shown in FIG. 1.

The air feeding system using one embodiment of the sheet feeding device of the present invention will be described in detail below with reference to the drawings. The air feeding system of this embodiment is characterized by the adjustment of control parameters used in air feeding, but first, the configuration of the entire air feeding system will be described. Figure 1 is a block diagram showing the schematic configuration of the air feeding system 100 of this embodiment. Figures 2 and 3 are diagrams showing the specific configuration of the feeder 1 in the air feeding system 100 shown in Figure 1. Also, Figure 4 is a block diagram showing the control system of the feeder 1. Note that the up, down, left, right, front, back directions shown in Figures 2 and 3 are the up, down, left, right, front, back directions of the feeder 1.

As shown in FIG. 1, the air sheet feeding system 100 of this embodiment includes a sheet feeding device 1 and a control device 2. The sheet feeding device 1 corresponds to the configuration of the sheet feeding device of the present invention other than the control parameter adjustment unit.

As shown in Figures 2 and 3, the paper feeder 1 includes a loading platform 10, an adsorption transport section 20, a suction mechanism 30, an imaging section 40, a transport mechanism 50, a transport sensor 51, a paper detection sensor 64, a main floating air supply section 71, a main separation air supply section 72, two side floating air supply sections 81, and two side separation air supply sections 82.

As shown in FIG. 4, the paper feeder 1 also includes a control unit 61, a memory unit 62, an interface unit 63, a loading platform lifting drive unit 91, a suction transport drive unit 92, a transport drive unit 93, and a shutter drive unit 94.

The loading tray 10 is used to load multiple sheets of paper P. The loading tray 10 is moved up and down by the loading tray lifting drive unit 91 shown in FIG. 4.

The suction conveying unit 20 has a conveying belt 21 and pulleys 22, 23 around which the conveying belt 21 is stretched. One of the pulleys 22, 23 is a driving pulley, and the other is a driven pulley. The driving pulley rotates by being driven by the suction conveying drive unit 92 shown in FIG. 4, causing the conveying belt 21 to rotate. As a result, the suction conveying unit 20 conveys the topmost sheet P1 in the conveying direction D (to the front in FIG. 2).

The conveyor belt 21 has multiple through holes for passing suction air A1 sucked in by the suction mechanism 30 described below.

Two suction transport units 20 are arranged side by side in the width direction of the paper P in the central area of the width direction (left and right direction in FIG. 2) perpendicular to the transport direction D of the paper P. In this case, it is preferable to arrange one suction mechanism 30 (described later) for each suction transport unit 20, but only one may be arranged.

The suction mechanism 30 sucks suction air A1 upward from the area surrounded by the conveyor belt 21 of the suction conveying section 20 through multiple through holes in the conveyor belt 21. As a result, the suction mechanism 30 causes the topmost sheet P1 of the multiple sheets P loaded on the stacking platform 10 to be adsorbed to the suction conveying section 20.

The transport mechanism 50 has a pair of rollers. One of the pair of rollers may be a drive roller and the other a driven roller, or both may be driven rollers. The drive roller of the transport mechanism 50 transports the paper P in the transport direction D1 by the transport drive unit 93 rotating the drive roller, and supplies it to the paper transport path of the printing device (not shown).

A transport sensor 51 is provided near the downstream side of the transport mechanism 50. The transport sensor 51 detects the paper P being transported by the transport mechanism 50. The transport sensor 51 detects whether the paper P has arrived at the correct timing.

The main floating air supply unit 71 is positioned downstream in the transport direction D from the multiple sheets of paper P loaded on the loading platform 10, and blows out main floating air A3 diagonally upward to float, for example, about 10 sheets of paper P including the topmost sheet P1.

The main floating air supply unit 71 includes a fan 71a and a shutter 31. The fan 71a generates the main floating air A3 by rotating. The shutter 31 switches between blocking and supplying the main floating air A3. Figures 5A and 5B are plan views showing the shutter 31. As shown in Figure 5A, the shutter 31 has a shutter body 31a, an opening member 31b, and a rotating shaft member 31c.

When driven by the shutter drive unit 94 shown in FIG. 4, the shutter body 31a swings (rotates) in both clockwise and counterclockwise directions within a range of, for example, 45 degrees, with the rotating shaft member 31c as the central axis of rotation.

The shutter body 31a has, for example, four blades 31a-1. These four blades 31a-1 are arranged at equal intervals (for example, 90 degree intervals) in the rotation direction of the shutter body 31a.

The opening member 31b is, for example, a disk-shaped plate material. The opening member 31b has, for example, four through holes 31b-1 to allow the main floating air A3 to pass through. The four through holes 31b-1 are provided at equal intervals (for example, 90 degree intervals) in the rotation direction of the shutter body 31a, similar to the blade portion 31a-1.

The shutter body 31a swings (moves) between a blocking position in which the blade portion 31a-1 covers the through hole 31b-1 to block the main floating air A3 shown in FIG. 5B, and a retracted position in which the blade portion 31a-1 does not cover the through hole 31b-1 to supply the main floating air A3 as shown in FIG. 5A.

The main separation air supply unit 72 is positioned downstream in the conveying direction D from the multiple sheets of paper P loaded on the loading platform 10, and blows out main separation air A4 diagonally upward to separate the top sheet of paper P1 from the second sheet of paper P2.

The main separation air supply unit 72 is equipped with a fan 72a. The fan 72a generates the main separation air A4 by rotating.

The side floating air supply units 81 are provided on the right and left sides in the width direction of the multiple sheets of paper P loaded on the stacking platform 10 (only the right side is shown in Figure 2), and similar to the main floating air supply unit 71, blow out side floating air A5 to float, for example, about 10 sheets of paper P including the topmost sheet of paper P1.

The side floating air supply unit 81 includes a fan 81a and a shutter 81b. The fan 81a generates the side floating air A5 by rotating. The shutter 81b has a configuration similar to that of the shutter 31, and is driven by the shutter drive unit 94 to switch between blocking and supplying the side floating air A5.

The side separation air supply units 82 are provided on the right and left sides in the width direction of the multiple sheets of paper P loaded on the loading platform 10, and blow out side separation air A6 to separate the top sheet of paper P1 from the second sheet of paper P2.

The side separation air supply unit 82 includes a fan 82a and a flow path 82b for the side separation air A6 supplied by the fan 82a.

The paper detection sensor 64 is a sensor that detects the passage of paper P, and has, for example, a transmissive optical sensor. The paper detection sensor 64 detects the arrival of paper P transported by the suction transport unit 20, and detects the thickness of the transported paper P to detect double feeding of paper P.

The photographing unit 40 photographs the end face of the downstream side (front side) in the transport direction of the paper P loaded on the loading platform 10, thereby photographing the floating state of the loaded paper P. The photographing unit 40 has an imaging element, such as a CMOS (Complementary Metal Oxide Semiconductor) image sensor or a CCD (Charge-Coupled Device) image sensor, and an imaging optical system. The photographing unit 40 is installed at a height that allows it to photograph the floating state of the loaded paper P and the downstream end of the suction transport unit 20, and has an angle of view.

In this embodiment, the image capturing unit 40 captures images at 60 fps, for example, from when a single sheet of paper P is adsorbed and transported from the stacking platform 10 until it is detected by the transport sensor 51, and outputs the captured images to the control device 2.

The control unit 61 controls the operation of the entire paper feeder 1. The control unit 61 includes a CPU (Central Processing Unit), semiconductor memory such as a ROM (Read Only Memory) and a RAM (Random access memory), storage such as a hard disk, and a communication I/F.

The control unit 61 controls the operation of each unit shown in FIG. 4 based on the control signal output from the control device 2. The control unit 61 also acquires detection signals from the paper detection sensor 64 and the transport sensor 51 and outputs them to the control device 2. In particular, the control unit 61 in this embodiment acquires control parameters of the main floating air supply unit 71 adjusted in the control device 2, and controls the rotation speed of the fan 71a of the main floating air supply unit 71 based on the acquired control parameters. In this embodiment, the duty ratio of the drive voltage of the fan 71a is acquired as the control parameter.

The loading platform lifting drive unit 91 has an actuator such as a motor that lifts and lowers the loading platform 10.

The suction conveying drive unit 92 has an actuator such as a motor that rotates a drive pulley, which is one of the pulleys 22, 23 of the suction conveying unit 20.

The transport drive unit 93 has an actuator such as a motor that rotates the drive roller of the transport mechanism 50.

The shutter driver 94 has an actuator such as a motor that rotates the shutter 31 and the shutter 81b.

Next, we will explain the operation of the paper feeder 1.

When paper feeding begins, the control unit 61 opens the shutter 31 to enable the supply of main floating air A3. The control unit 61 also starts driving the suction mechanism 30, the main floating air supply unit 71, the main separation air supply unit 72, the side floating air supply unit 81, and the side separation air supply unit 82.

As a result, main floating air A3 and side floating air A5 are blown out from the main floating air supply section 71 and side floating air supply section 81, and main separation air A4 and side separation air A6 are blown out from the main separation air supply section 72 and side separation air supply section 82. In addition, an adhesive force is generated at the through holes of the conveyor belt 21.

The floating air currents of the main floating air A3 and the side floating air A5 cause the uppermost sheets P of the paper stack to float from the paper stack, and the uppermost sheet P of these is adsorbed and held on the transport surface of the transport belt 21.

On the other hand, the main separation air A4 and the side separation air A6 are sent between the topmost sheet P and the second sheet P from the top, and flow upstream.

In this state, the control unit 61 closes the shutter 31. This stops the blowing of the floating air current of the main floating air A3. As a result, the topmost sheet P is separated from the second and subsequent sheets P by the main separation air A4 and the side separation air A6, and the second and subsequent sheets P from the top fall.

Next, the control unit 61 drives the suction transport drive unit 92 to cause the suction transport unit 20 to transport the paper P.

Next, the control unit 61 opens the shutter 31 to raise the paper P so that the next paper P can be fed.

By repeating the above operations, paper P is fed sequentially from the paper feeder 1 to the printing device.

During this paper feeding operation, the control unit 61 controls the stacking tray 10 to rise in response to the reduction in the amount of paper P on the stacking tray 10 due to paper feeding.

Next, we will explain the control device 2.

The control device 2 acquires the image captured by the imaging unit 40 of the paper feeder 1, adjusts the control parameters of the main floating air supply unit 71 (in this embodiment, the duty ratio of the drive voltage of the fan 71a) based on the captured image, and outputs the control parameters to the control unit 61 of the paper feeder 1. The control unit 61 of the paper feeder 1 controls the main floating air supply unit 71 based on the control parameters output from the control device 2.

Specifically, the control device 2 includes an image recognition unit 80 and a control parameter adjustment unit 83.

The image recognition unit 80 acquires the captured image captured by the imaging unit 40 of the paper feeder 1, and acquires information on the floating state of the paper P in the paper feeder 1 based on the captured image. In this embodiment, the image recognition unit 80 acquires, as floating state information, the position of the floated paper P, the number of floated sheets of paper P, the adsorption state to the adsorption transport unit 20, the occurrence state of double-layer floating and triple-layer floating, the shaking state of the paper P, and the scattered state of the paper P. It is not necessary to acquire all of this floating state information, and it is sufficient to acquire at least one of them.

The image recognition unit 80 detects the edge of the paper P from the captured image.

The image recognition unit 80 obtains information on the position of the floated paper P, which is the position of the second sheet of paper P following the topmost sheet of paper P adsorbed by the adsorption transport unit 20.

Specifically, the position of the second sheet of paper P described above is detected from multiple images captured while one sheet of paper P is being passed through, and the average value of these positions is obtained as information on the position of the floated sheet of paper P.

The information on the number of floated sheets of paper P is obtained by averaging the number of floated sheets of paper P detected from each of the multiple captured images described above.

As the state of adsorption to the adsorption conveying section 20, information is acquired indicating whether the topmost sheet P is adsorbed to the adsorption conveying section 20. In this embodiment, whether the sheet P is adsorbed to the adsorption conveying section 20 is detected by detecting a through hole in the conveying belt 21 from a captured image of the downstream end of the adsorption conveying section 20. Specifically, if a through hole in the conveying belt 21 is detected from the captured image, information is acquired indicating that the sheet P is not adsorbed to the adsorption conveying section 20, and if a through hole in the conveying belt 21 is not detected from the captured image, i.e., if the through hole is blocked by the sheet P, information is acquired indicating that the sheet P is adsorbed to the adsorption conveying section 20.

The occurrence status of double-layer floating and triple-layer floating is obtained from the captured images. Specifically, the number of detections of locations where double-layer floating is occurring and the total number of detections of locations where double-layer floating is occurring are obtained.

The shaking state of the paper P refers to the flapping state of the multiple sheets of paper P stacked in stack number 10. The flapping of the paper P refers to the flapping of each sheet of paper P that has been lifted by the floating air, and if the flapping is large, the captured image will be blurred, making it impossible to detect the edge of each sheet of paper P.

The image recognition unit 80 obtains information on the blur state of the paper P by determining whether or not there is a captured image in which the edge of each paper P cannot be detected due to blurring as described above.

The dispersion state of the paper P refers to the degree to which the multiple sheets of paper P that have been floated by the floating air are dispersed. Specifically, the average sheet spacing and the variation in sheet spacing are obtained as the dispersion state of the paper P.

The control parameter adjustment unit 83 adjusts the control parameters of the main floating air supply unit 71 based on the above-mentioned floating state information acquired from the captured image by the image recognition unit 80, the information on the detection of double feed by the paper detection sensor 64, the arrival information of the paper P at the transport sensor 51, and the paper quality information of the paper P. Note that in this embodiment, as described above, the control parameter is the duty ratio of the drive voltage that drives the fan 71a of the main floating air supply unit 71.

As the information on the double feed detection, information on whether or not a double feed has been detected is acquired. As the information on the arrival of the paper P, information on whether the timing at which the paper P was detected by the transport sensor 51 was earlier (too early) or later (not yet arrived) than the normal timing is acquired. As the paper quality information of the paper P, information on the size of the paper P and the paper type of the paper P is acquired. As the paper quality information of the paper P, information set and input by the user, for example, is acquired. Also, the paper quality information of the paper P may be acquired from the print job.

The control parameter adjustment unit 83 stores a learning model that has been obtained in advance by reinforcement learning of the relationship between the predetermined control parameters of the main floating air supply unit 71, floating state information obtained from the captured image when paper is passed using the control parameters, the paper passing results including the double feed information and the arrival information of the paper P, and the paper quality information of the paper P. Note that the floating state information obtained from the captured image, the paper passing results, and the paper quality information of the paper P are environmental parameters of the reinforcement learning, and are shown together in FIG. 6.

When performing reinforcement learning, a reward is given to the learning model. The reward is determined by determining whether the environmental parameters shown in FIG. 7 satisfy the reward plus conditions. If all the reward plus conditions are satisfied, a positive reward (for example, +10) is given to the learning model. If any of the reward plus conditions are not satisfied, a negative reward (for example, -1) is given to the learning model. Note that the non-detection of the presence or absence of an adhesion state shown in FIG. 7 means that the non-adhesion state is not detected (the abnormal state is not detected). When a negative reward is given, the learning model outputs a control parameter that takes into account the adjustment amount as shown in FIG. 8. When a positive reward is given, the learning model outputs the same control parameter as the input control parameter. Then, the control parameter that takes into account the adjustment amount is used to perform the next paper feed, and the same process as above is repeated to proceed with reinforcement learning and generate a learning model.

Next, we will explain how to adjust the control parameters of the paper feeder 1 using the learning model described above, with reference to the flowchart shown in Figure 9.

First, when a user inputs a command to start printing, the paper feeder 1 operates as described above, the paper P rises, and paper feeding begins (S10). At this time, preset initial values are used as the control parameters.

Then, the floating state of the paper P is photographed by the photographing unit 40 of the paper feeder 1 (S12), and the photographed image is input to the image recognition unit 80 of the control device 2, and information on the floating state is obtained from the photographed image by the image recognition unit 80. Then, after the paper is passed through, the information on the floating state, the paper quality information of the paper P, and the information on the paper passing result are input to the control parameter adjustment unit 83 as environmental parameters (S14).

The control parameter adjustment unit 83 inputs the input environmental parameters into the learning model, obtains the adjustment amount of the control parameters (S16), and outputs the control parameters that take into account the adjustment amount to the paper feed device 1.

The control parameter adjustment unit 83 also obtains a reward based on the environmental parameters (S18). If the reward is negative (S20, YEN) or if the reward is positive but this is the first time the reward has been positive (S22, NO), the control parameter adjustment unit 83 continues to adjust the control parameters by repeating the processes of S12 to S18 after the next paper feed.

On the other hand, if a positive reward is obtained twice in a row (S20, NO; S22, YES), the control parameter adjustment unit 83 ends the control parameter adjustment process (S24), and the paper feeder 1 continues to use the same control parameters to feed paper.

In this embodiment, as described above, the adjustment process of the control parameters is terminated when two consecutive positive rewards are obtained. However, if two consecutive positive rewards are not obtained even after performing the adjustment process ten times, for example, the adjustment process may be terminated after ten times.

In addition, in this embodiment, the drive voltage of the main floating air supply unit 71 is adjusted as a control parameter, but this is not limited to this, and other factors that affect the floating state of the paper P, such as the opening and closing timing of the shutter 31, may also be included in the control parameters.

The control device 2 includes a CPU, semiconductor memory such as ROM and RAM, storage such as a hard disk, and a communication I/F. An embodiment of the sheet supply program of the present invention is installed in the semiconductor memory or hard disk of the control device 2, and the sheet supply program is executed by the CPU to perform the above-mentioned control parameter adjustment process.

In this embodiment, the control parameter adjustment process is realized by the sheet supply program, but this is not limited to the above, and some or all of the functions or controls may be realized by hardware such as an ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or other electrical circuits.

Figure 10 is a diagram showing an example of the air paper feeding system 100 of the above embodiment. In the example shown in Figure 10, the vertical axis represents the duty ratio of the drive voltage supplied to the main floating air supply unit 71, and the horizontal axis represents the number of adjustment processes described above. Figure 10 also shows the results of adjusting the control parameters (duty ratio of the drive voltage) for four types of paper P. The dotted line range shown in Figure 10 indicates the optimal control parameter range, and the vertical axis represents the deviation from the optimal control parameter range. As shown in Figure 10, it can be seen that the control parameters converge to the optimal range each time the adjustment process using the learning model is performed multiple times.

According to the air paper feed system 100 of the above embodiment, the floating state of multiple sheets due to the supply of air is photographed, and the control parameters of the main floating air supply unit 71 are adjusted based on the information on the floating state obtained from the photographed image. This makes it possible to adjust the control parameters more optimally, taking into account the actual floating state of the paper P, and prevents empty feeding and double feeding.

In addition, the information on the floating state is obtained by acquiring at least one of the following: the position of the floating paper P, the number of floating sheets, the adsorption state to the adsorption transport section, the blurring state of the paper P, and the scattered state of the sheets, so that the control parameters can be adjusted with higher accuracy.

In addition, the control parameters are adjusted based on the floating state information obtained from the captured image and the paper passing results, so the control parameters can be adjusted to reflect the paper passing results.

In addition, the control parameters are adjusted based on the floating state information obtained from the captured image, the paper quality information of the paper P, and the paper passing results, so the control parameters can be adjusted according to the paper quality information of the paper P.

In addition, the control parameters are adjusted using a learning model obtained by performing reinforcement learning on the relationship between predetermined control parameters and information on the floating state obtained from images captured when paper is passed using those control parameters. This means that the control parameters are adjusted automatically, eliminating the need for special skills to make the adjustments.

In addition, the learning model is trained using rewards determined based on at least one of the floating state information and the paper passing results, which reduces the time required to adjust the control parameters and reduces the downtime of the device.

In addition, if positive rewards are obtained consecutively with the same control parameters, adjustment of the control parameters is terminated, allowing adjustment of the control parameters to be performed more efficiently.

The present invention is not limited to the above embodiment, and can be embodied by modifying the components without departing from the spirit of the invention. Various inventions can be formed by appropriately combining the multiple components disclosed in the above embodiment. For example, all the components shown in the embodiment can be appropriately combined. It goes without saying that various modifications and applications are possible without departing from the spirit of the invention.

The following additional notes are further disclosed regarding the present invention:

(Appendix 1)
The sheet supplying device of the present invention comprises a loading platform on which multiple sheets are loaded, a floating air supply unit that floats the multiple sheets loaded on the loading platform by supplying air toward the end faces of the multiple sheets, an adsorption and transport unit that adsorbs and transports the uppermost sheet of the multiple sheets floated by the supply of air, a photographing unit that photographs the floating state of the multiple sheets by the supply of air, a control parameter adjustment unit that adjusts control parameters of the floating air supply unit based on information about the floating state obtained from the image captured by the photographing unit, and a control unit that controls the floating air supply unit based on the control parameters adjusted by the control parameter adjustment unit.

(Appendix 2)
In the sheet supplying device described in Appendix 1, the control parameter adjustment unit can obtain at least one of the following information on the floating state: the position of the floating sheet, the number of floating sheets, the adsorption state to the adsorption conveying unit, the sheet vibration state, and the sheet scattering state.

(Appendix 3)
In the sheet supplying device according to claim 1 or 2, the control parameter adjustment unit can adjust the control parameters based on the floating state information acquired from the captured image and the sheet passing result.

(Appendix 4)
In the sheet supplying device described in any one of Supplementary Note 1 to Supplementary Note 3, the control parameter adjustment unit can adjust the control parameters based on information on the floating state acquired from the captured image, paper quality information of the sheet, and the sheet passing result.

(Appendix 5)
In the sheet feeding device described in any one of Supplementary Notes 1 to 4, the control parameter adjustment unit can adjust the control parameters by using a learning model obtained in advance by reinforcement learning of the relationship between a predetermined control parameter and information on the floating state obtained from an image captured when paper is passed using the control parameter.

(Appendix 6)
In the sheet feeding device according to Supplementary Note 5, the learning model is preferably trained by a reward determined based on at least one of the levitation state and the sheet threading result.

(Appendix 7)
In the sheet feeding device according to Supplementary Note 6, it is preferable that the control parameter adjustment unit terminates adjustment of the control parameter when a positive reward is consecutively obtained by the same control parameter.

(Appendix 8)
The sheet supplying method of the present invention involves supplying air toward the end faces of multiple sheets stacked on a stacking table to float the sheets, and then adsorbing and transporting the topmost sheet of the floated multiple sheets.The method photographs the floating state of the multiple sheets caused by the air supply, and based on information about the floating state obtained from the photographed image, adjusts control parameters of a floating air supply unit that supplies air, and controls the floating air supply unit based on the adjusted control parameters.

(Appendix 9)
The sheet supply program of the present invention causes a computer to execute the following steps when floating the sheets by supplying air toward the end faces of multiple sheets stacked on a stacking table, and adsorbing and transporting the topmost sheet of the floated multiple sheets: photographing the floating state of the multiple sheets by supplying air; adjusting control parameters of a floating air supply unit that supplies air based on information about the floating state obtained from the photographed image; and controlling the floating air supply unit based on the adjusted control parameters.

(Appendix 10)
The learning model of the present invention is a learning model used when adjusting the control parameters of the floating air supply unit of the sheet feeding device described in Appendix 1, and is obtained by prior reinforcement learning of the relationship between a predetermined control parameter and information on the floating state obtained from an image captured by an imaging unit of the sheet feeding device when paper is passed using that control parameter.

1 Paper feeder 2 Control device 10 Loading platform 20 Suction conveying section 21 Conveyor belt 22, 23 Pulley 30 Suction mechanism 31 Shutter 31a Shutter body 31a-1 Blade section 31b Opening member 31b-1 Through hole 31c Rotating shaft member 40 Photography section 50 Conveying mechanism 51 Conveying sensor 61 Control section 62 Memory section 63 Interface section 64 Paper detection sensor 70 Control parameter adjustment section 71 Main floating air supply section 71a Fan 72 Main separation air supply section 72a Fan 80 Image recognition section 81 Side floating air supply section 81a Fan 81b Shutter 82 Side separation air supply section 82a Fan 82b Flow path 83 Control parameter adjustment section 91 Loading platform lifting and lowering drive section 92 Suction conveying drive section 93 Conveying drive section 94 Shutter drive section 100 Air paper feeding system

Claims (10)

A loading platform on which a plurality of sheets are loaded;
a floating air supply unit that floats the sheets by supplying air toward end faces of the sheets stacked on the stacking table;
a suction conveying section that suctions and conveys the uppermost sheet of the plurality of sheets floated by the supply of air;
an image capturing unit that captures an image of a floating state of the plurality of sheets due to the supply of air;
a control parameter adjustment unit that adjusts a control parameter of the floating air supply unit based on information about the floating state obtained from the captured image by the photographing unit;
a control unit that controls the floating air supply unit based on the control parameters adjusted by the control parameter adjustment unit. The sheet supply device according to claim 1, wherein the control parameter adjustment unit acquires at least one of the following information on the floating state: the position of the floating sheet, the number of floating sheets, the adsorption state to the adsorption transport unit, the sheet vibration state, and the sheet scattering state. The sheet supplying device according to claim 1, wherein the control parameter adjustment unit adjusts the control parameters based on information on the floating state obtained from the captured image and the paper passing results. The sheet supplying device according to claim 3, wherein the control parameter adjustment unit adjusts the control parameters based on information on the floating state obtained from the captured image, paper quality information of the sheet, and the paper passing result. The sheet feeder according to claim 1, wherein the control parameter adjustment unit adjusts the control parameters using a learning model obtained by performing reinforcement learning on the relationship between a predetermined control parameter and information on the floating state obtained from the captured image when paper is passed using the control parameter. The sheet feeder of claim 5, wherein the learning model is trained with a reward determined based on at least one of the information on the floating state and the paper passing result. The sheet supply device according to claim 6, wherein the control parameter adjustment unit terminates adjustment of the control parameter when a positive reward is obtained consecutively with the same control parameter. 1. A sheet supplying method for supplying air toward end faces of a plurality of sheets stacked on a stacking table to float the sheets, and then adsorbing and conveying a topmost sheet of the plurality of floated sheets, comprising:
The sheet supplying method includes photographing the floating state of the multiple sheets due to the supply of air, adjusting control parameters of a floating air supply unit that supplies the air based on information about the floating state obtained from the photographed image, and controlling the floating air supply unit based on the adjusted control parameters. A sheet supply program that causes a computer to execute the steps of: supplying air toward the end faces of multiple sheets loaded on a loading platform to float the sheets; and, when adsorbing and transporting the topmost sheet of the multiple floated sheets, photographing the floating state of the multiple sheets caused by the supply of air; adjusting control parameters of a floating air supply unit that supplies the air based on information about the floating state obtained from the photographed image; and controlling the floating air supply unit based on the adjusted control parameters. A learning model used when adjusting a control parameter of a floating air supply unit of the sheet feeding device according to claim 1,
A learning model obtained by prior reinforcement learning of the relationship between a specified control parameter and information on the floating state obtained from an image captured by an imaging unit of the sheet feeding device when paper is passed using the control parameter.