阅读说明：本技术 树脂膜制造装置及树脂膜制造方法 (Resin film manufacturing apparatus and resin film manufacturing method ) 是由 前西隆一郎 原本信洋 藤井文武 于 2020-03-09 设计创作，主要内容包括：本发明涉及一种树脂膜制造装置及树脂膜制造方法,在一种实施方式的树脂膜制造装置中,首先根据从厚度传感器获得的从树脂膜厚度分布中计算出的控制误差,确定每一加热螺栓的当前状态以及针对之前选择的操作的回馈,然后根据该回馈,更新作为状态和条件的组合的控制条件,并从更新后的控制条件中选择与当前状态相应的最恰当操作。随后,根据所述最恰当操作,对加热件进行控制。(In one embodiment, a current state of each heating bolt and a feedback for a previously selected operation are first determined based on a control error calculated from a resin film thickness distribution obtained from a thickness sensor, and then, based on the feedback, a control condition that is a combination of the state and the condition is updated, and an optimum operation corresponding to the current state is selected from the updated control conditions. Subsequently, the heating member is controlled in accordance with the most appropriate operation.)

1. A resin film manufacturing apparatus, comprising:

a die including a plurality of pairs of heating bolts arranged along long sides of a pair of die lips and a heating member for heating the heating bolts, the die being capable of adjusting a die lip gap for each of the heating bolts;

a cooling roller that cools the film-shaped molten resin extruded from the gap between the pair of die lips and discharges a resin film that is a solidified form of the molten resin;

a thickness sensor that measures a thickness distribution in a width direction of the resin film discharged from the cooling roll; and

a control unit that performs feedback control on the die lip gap based on a thickness distribution obtained from the thickness sensor, wherein,

for each of the heating bolts, the control unit:

determining a current state and feedback for a previously selected operation based on a control error calculated from the thickness profile;

updating control conditions according to the feedback, and selecting the most appropriate operation corresponding to the current state from the updated control conditions, wherein the control conditions are the combination of the state and the operation; and is

Controlling the heating member according to the most suitable operation.

2. The resin film manufacturing apparatus according to claim 1, wherein said operation is modification of an output of said heating member.

3. The resin film manufacturing apparatus according to claim 1, wherein said operation is modification of a parameter of a PID controller for controlling an output of said heating member.

4. The resin film manufacturing apparatus according to claim 1, wherein the thickness sensor is a non-contact sensor.

5. The resin film manufacturing apparatus according to claim 4, wherein the thickness sensor measures a thickness distribution in a width direction of the resin film by scanning in the width direction of the resin film.

6. The resin film manufacturing apparatus according to claim 5, wherein the thickness sensor measures the thickness distribution in the width direction of the resin film that is horizontally conveyed.

7. The resin film manufacturing apparatus according to claim 1, wherein only one of the pair of die lips is connected to the heating bolt.

8. A method for producing a resin film, comprising the steps of:

(a) extruding a film-like molten resin from a gap between a pair of die lips of a die;

(b) conveying a resin film, which is a solidified form of the molten resin, and measuring a thickness distribution of the resin film in a width direction; and

(c) feedback control is applied to the die lip gap based on the measured thickness profile, wherein,

the die includes a plurality of pairs of heating bolts provided along the long sides of the pair of die lips and a heating member for heating the heating bolts, and the die lip gap can be adjusted for each of the heating bolts,

in the step (c), for each of the heating bolts, a computer:

(c1) determining a current state and feedback for a previously selected operation based on a control error calculated from the thickness profile;

(c2) updating control conditions according to the feedback, and selecting the most appropriate operation corresponding to the current state from the updated control conditions, wherein the control conditions are the combination of the state and the operation; and is

(c3) Controlling the heating member according to the most suitable operation.

9. The resin film manufacturing method according to claim 8, wherein said operation determined in said step (c2) is modification of an output of said heating element.

10. The resin film manufacturing method according to claim 8, wherein said operation determined in said step (c2) is modification of a parameter of a PID controller for controlling an output of said heating element.

11. The resin film manufacturing method according to claim 8, wherein in the step (b), a thickness distribution in a width direction of the resin film is measured by a noncontact thickness sensor.

12. The resin film manufacturing method according to claim 11, wherein the thickness sensor measures a thickness distribution in a width direction of the resin film by scanning in the width direction of the resin film.

13. The resin film manufacturing method according to claim 12, wherein the thickness sensor measures the thickness distribution in the width direction of the resin film that is horizontally conveyed.

14. The resin film manufacturing method according to claim 8, wherein only one of the pair of die lips is attached to the heating bolt.

Technical Field

The present invention relates to a resin film manufacturing apparatus and a resin film manufacturing method.

Background

In one known resin film manufacturing apparatus, a film-shaped molten resin is extruded through a gap between die lips provided on an extruder. In such a resin film manufacturing apparatus, it is necessary to achieve a uniform film thickness in the resin film width direction.

Therefore, the mold disclosed in japanese unexamined patent application publication nos. 2010-167584, 2012-240332, 2013-052574 includes a plurality of heating bolts provided along the long side direction of the die lip (the resin film width direction). By individually adjusting the degree of thermal expansion caused by the heater of each heating bolt, the die lip gap of the die can be locally adjusted.

Further, japanese unexamined patent application publication No. 2013-039677 discloses a resin film manufacturing apparatus capable of measuring the thickness of a resin film during the manufacturing process and performing feedback control on the die lip gap of a mold.

Disclosure of Invention

The present inventors have found that various problems are involved in the development of a resin film manufacturing apparatus including a die having a plurality of heating bolts and capable of feedback-controlling a lip gap.

Other problems and novel features of the present disclosure will become apparent from the description of the specification and the drawings.

In the resin film manufacturing apparatus of an embodiment, first, the current state of each heating bolt and the feedback for the previously selected operation are determined based on the control error calculated from the resin film thickness distribution obtained from the thickness sensor, then, based on the feedback, the control conditions as the state/condition combination are updated, and the most appropriate operation corresponding to the current state is selected from the updated control conditions. Subsequently, the heating member is controlled in accordance with the most appropriate operation.

According to this embodiment, an excellent resin film manufacturing apparatus can be provided.

The above and other objects, features and advantages of the present disclosure will be more fully understood from the following detailed description and the accompanying drawings. Wherein the description and drawings are for illustrative purposes only and are not to be construed as limiting the present disclosure.

Drawings

Fig. 1 is a schematic cross-sectional view of the overall structure of a resin film manufacturing apparatus and a resin film manufacturing method according to a first embodiment.

Fig. 2 is a cross-sectional view of the T-die 20.

Fig. 3 is a partial perspective view of the lower portion (with die lips) of the T-die 20.

Fig. 4 is a block diagram showing the configuration of the control unit 70 according to the first embodiment.

Fig. 5 is a flowchart of a method for controlling a die lip gap in the method for manufacturing a resin film according to the first embodiment.

Fig. 6 is a block diagram showing the structure of the control unit 70 according to the second embodiment.

Detailed Description

Hereinafter, specific embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. However, the present disclosure is not limited to the following embodiments. The following description and drawings are appropriately shortened and simplified for clarity of illustration.

First embodiment

< integral Structure of resin film production apparatus >

First, referring to fig. 1, the overall structure of the resin film manufacturing apparatus and the resin film manufacturing method of the first embodiment is described. Fig. 1 is a schematic cross-sectional view of the overall structure of a resin film manufacturing apparatus and a resin film manufacturing method according to a first embodiment.

It should be noted that the right-handed spiral xyz cartesian coordinate system is given in fig. 1 and other figures for the purpose of facilitating the description of the positional relationship between the elements. In general, in all the drawings, the Z-axis forward direction is the vertically upward direction, and the XY plane is the horizontal plane.

Further, in the present specification, the resin film includes a resin sheet.

As shown in fig. 1, the first embodiment resin film manufacturing apparatus includes an extruder 10, a T-die 20, a cooling roll 30, a conveying roll 40, a winder 50, a thickness sensor 60, and a control unit 70. The first embodiment resin film manufacturing apparatus is an extrusion molding resin film manufacturing apparatus that extrudes a film-like molten resin 82a through a gap between die lips of a T-die 20 provided on an extruder 10.

The extruder 10 may be a screw extruder, for example. In the extruder 10 shown in fig. 1, a screw 12 disposed along the X-axis is housed in a cylinder 11 disposed along the X-axis. A hopper 13 is provided on the X-axis negative side edge of the cylinder 11, and resin pellets 81 as a raw material for a resin film 83 are placed therein.

The resin pellets 81 fed through the hopper 13 are extruded from the root of the rotating screw 12 to the X-axis direction one end portion thereof. The rotating screw 12 in the cylinder 11 extrudes the resin pellets 81 into molten resin 82.

It is to be noted that, although not shown in the drawings, for example, a motor as a power source is connected to the screw 12 through a reduction gear.

As shown in FIG. 1, a T-die 20 is provided below the end of the extruder 10 (the X-axis forward side edge). The film-like molten resin 82a is extruded downward (in the negative Z-axis direction) through the gap between the die lips provided at the lower end of the T-die 20. The lip gap of the T-die 20 can be adjusted. Specifically, the die lip gap of the T-die 20 can be adjusted at a plurality of positions in the long side direction (Y-axis direction) of the die lip so that the resin film 83 to be produced has a uniform thickness in the width direction (Y-axis direction) thereof, which will be described in detail below.

The cooling roll 30 is used to cool the film-shaped molten resin 82a extruded from the T-die 20 and output the resin film 83 in a solidified form as the film-shaped molten resin 82 a. The resin film 83 output from the cooling roll 30 is conveyed to the winder 50 by the conveying roll 40, and is wound by the winder 50. In the example of fig. 1, the conveying roller 40 includes eight conveying rollers 41 to 48. The number and position of the conveying rollers can be set as appropriate.

The thickness sensor 60 is a non-contact type thickness sensor, and measures, for example, the thickness distribution of the resin film 83 discharged and transferred by the cooling roll 30 in the width direction thereof. In the example of fig. 1, the thickness sensor 60 is disposed such that the resin film 83 horizontally conveyed between the conveying rollers 44 and 45 is between the upper and lower portions of the sensor. Since the thickness sensor 60 is a non-contact sensor, it can scan in the width direction (Y-axis direction) of the resin film 83. In this manner, the thickness distribution in the width direction of the resin film 83 is allowed to be measured using the compact thickness sensor 60. Further, since the resin film 83 is horizontally conveyed, the thickness distribution can be accurately measured by scanning of the thickness sensor 60.

The control unit 70 performs feedback control of the lip gap of the T-die 20 based on the thickness distribution of the resin film 83 obtained by the thickness sensor 60. Specifically, the control unit 70 allows the resin film 83 to obtain a uniform thickness along its width by controlling the lip gap of the T-shaped die 20. Hereinafter, the structure and operation of the control unit 70 will be described in detail.

< Structure of T-shaped mold 20 >

Referring to fig. 2 and 3, the structure of the T-shaped mold 20 is described in further detail. Fig. 2 is a cross-sectional view of the T-die 20. Fig. 3 is a partial oblique view of the lower portion (with die lips) of the T-die 20.

As shown in fig. 2 and 3, the T-die 20 is composed of a pair of die blocks 21 and 22 abutting against each other. Each of the pair of mutually abutting die blocks 21 and 22 has a tapered portion whose outer surface is inclined downward toward the inner surface (facing surface). In this manner, the die pieces 21 and 22 are respectively provided with die lips 21a and 22a having a small thickness at the lower ends of the facing surfaces thereof.

The facing surfaces of the pair of die blocks 21 and 22 are provided with a feed opening 20a, a manifold structure 20b and a slit 20 c. The feed port 20a extends downward (in the negative Z-axis direction) from the upper surface of the T-die 20. The manifold structure 20b extends from the lower end of the feed port 20a toward the positive Y-axis direction and the negative Y-axis direction. In this manner, the feed throat 20a and the manifold structure 20b together form a T-shape within the T-die 20.

Further, the slit 20c extends from the bottom of the manifold structure 20b to the lower surface of the T-shaped die 20 in the Y-axis direction. Molten resin 82 is extruded downward from slit 20c (i.e., the gap between lips 21a and 22 a) through throat 20a and manifold structure 20 b.

When the die lip 21a is a fixed die lip which is not movable, the die lip 22a is a movable die lip connected to the heating bolt 23. The die lip 22a has a cut groove 22b cut obliquely upward from its outer surface toward the facing surface. The die lip 22a is pushed and pulled by the heating bolt 23 and is thus movable with respect to the bottom of the cut groove 22 b. Since only the die lip 22a is a movable die lip, adjustment of the die lip gap can be easily achieved with a simple structure.

The heating bolts 23 extend obliquely upward along the tapered portion of the die block 22. The heating bolts 23 are supported by holders 25a and 25b fixed to the die block 22. More specifically, the heating bolt 23 is screwed into the screw hole of the holder 25 a. The tightening degree of the heating bolt 23 can be appropriately adjusted. When the heating bolt 23 passes through the through-hole of the holder 25b, it is not fixed to the holder 25 b. It should be noted that the retainers 25a and 25b do not necessarily have to be separate from the block 22, but may be formed integrally with the block 22.

As shown in fig. 3, a plurality of heating bolts 23 are provided along the longitudinal direction (Y-axis direction) of the die lips 21a and 22 a. The long side direction of the die lips 21a and 22a corresponds to the width direction of the resin film. Although three heating bolts 23 are schematically shown in fig. 3, more heating bolts 23 are generally provided.

Each heating bolt 23 is provided with a heating element 24 to heat the heating bolt 23. In the example of fig. 2 and 3, the heating element 24 of each heating bolt 23 is arranged to cover its outer surface between the holders 25a and 25 b. By tightening the heating bolt 23, the lower end face of the heating bolt 23 will exert a thrust force against the die lip 22 a. The lower end surface of the heating bolt 23 is connected to the die lip 22a by a connecting member 26 having a U-shaped cross section and fixed to the die lip 22 a. Thus, by loosening the heating bolts 23, the connecting member 26 will apply a tensile force to the die lip 22 a.

The distance between the lips 21a and 22a can be adjusted by the tightening degree of the heating bolts 23. Specifically, when the tightening degree of the heating bolt 23 is increased, the heating bolt 23 pushes the die lip 22a, so that the gap between the die lips 21a and 22a is narrowed. In contrast, when the tightening degree of the heating bolt 23 is reduced, the gap between the die lips 21a and 22a will be widened. The tightening degree of the heating bolt 23 is adjusted, for example, manually.

Further, the gap between the die lips 21a and 22a can also be finely adjusted by thermal expansion of the heating bolt 23 caused by the heating member 24. Specifically, when the heating temperature of the heating member 24 is raised, the degree of thermal expansion of the heating bolt 23 becomes large, so that the heating bolt 23 pushes the die lip 22a and the gap between the die lips 21a and 22a becomes narrow. In contrast, when the heating temperature of the heating member 24 is lowered, the degree of thermal expansion of the heating bolt 23 becomes small, thereby increasing the gap between the die lips 21a and 22 a. The degree of thermal expansion of each heating bolt 23, i.e., the heating of each heating member 24, is controlled by the control unit 70.

< construction of control means 70 of comparative example >

The overall structure of the resin film manufacturing apparatus of the comparative example is similar to that of the resin film manufacturing apparatus of the first embodiment shown in fig. 1. In the comparative example, the control unit performs feedback control on the heating member 24 of each heating bolt 23 based on the thickness distribution of the resin film 83 obtained by the thickness sensor 60 by using PID control by the control unit 70. In PID control, the parameters need to be adjusted each time the process conditions are changed. In general, an operator performs parameter adjustment according to a trial and error method, and thus the parameter adjustment work requires a lot of time and a lot of resin materials.

< construction of control means 70 of first embodiment >

Hereinafter, the structure of the control unit 70 of the first embodiment will be described in more detail with reference to fig. 4. Fig. 4 is a block diagram showing the configuration of the control unit 70 according to the first embodiment. As shown in fig. 4, the first embodiment control unit 70 includes a state observation unit 71, a control condition learning unit 72, a storage unit 73, and a control signal output unit 74.

Note that the functional blocks of the control unit 70 may be implemented by a CPU (central processing unit), a memory, or other circuit configuration as hardware, or a program loaded into a memory or the like as software. It can thus be seen that these functional modules may be implemented in various ways by computer hardware, software, or a combination thereof.

The state observation unit 71 calculates a control error of each heating bolt 23 based on the measured value pv of the thickness distribution of the resin film 83 obtained by the thickness sensor 60. The control error is the difference between the target value and the measured value pv. The target value is an average value of the measured values pv of the thickness distribution of the resin film 83 of all the heating bolts 23 measured by the thickness sensor 60.

It is to be noted that, in calculating the average value of the measured values pv, the measured values at both ends of the resin film 83 which is not used as a product may be excluded.

Further, the measured value pv of each heating bolt 23 is obtained from the measured value pv of the thickness at the specified measurement point of the heating bolt 23. For example, the measured value pv of each heating bolt 23 is an average value of the thickness measured values pv at the specified measurement points of the heating bolt 23. Alternatively, the measured value pv of each heating bolt 23 may also specify, for that heating bolt 23, the measured value pv of the thickness that differs the greatest from the target value at the measurement point.

Subsequently, the state observing unit 71 determines the current state st of each heating bolt 23 and the feedback rw for the previously (e.g., last) selected operation ac based on the calculated control error.

The state st is predetermined to divide the control error values into a finite number of groups, which may be an infinite number of values. For illustrative purposes, only a simple example is illustrated here: when the control error is err, the state st1 is defined as-0.9 μm-err < -0.6 μm, -0.6 μm-err < -0.3 μm is defined as the state st2, -0.3 μm-err <0.3 μm is defined as the state st3, 0.3 μm-err <0.6 μm is defined as the state st4, and 0.6 μm-err < 0.9 μm is defined as the state st5, and so on. In practical applications, a larger number of more subdivided states st are often set.

The feedback rw is an index evaluating the operation ac selected in the previous state st.

Specifically, when the absolute value of the calculated current control error value is smaller than the absolute value of the previous control error, the state observation unit 71 determines the previously selected operation ac as the appropriate operation, and sets the feedback rw to a positive value, for example. In other words, the feedback rw is determined so that the operation ac selected previously may be selected again in the same state as before.

In contrast, when the absolute value of the calculated current control error value is larger than the absolute value of the previous control error, the state observation unit 71 determines that the previously selected operation ac is inappropriate, and sets the feedback rw to a negative value, for example. In other words, the feedback rw is determined so that the operation ac selected previously cannot be selected again in the same state as before.

Specific examples of feedback rw will be given later. The value of feedback rw can be determined as appropriate. For example, the value of feedback rw may always be positive; alternatively, the value of feedback rw may always be negative.

The control condition learning unit 72 performs reinforcement learning for each heating bolt 23. Specifically, the control condition learning unit 72 updates the control conditions (learning results) based on the feedback rw, and selects the most appropriate operation ac corresponding to the current state st from the updated control conditions. The control condition is a combination of the state st and the operation ac. Table 1 shows simple control conditions (learning results) corresponding to the states st1 to st 5. In the example of fig. 4, the control condition learning unit 72 stores the updated control condition cc in the storage unit 73 (e.g., memory), and updates the control condition cc after reading it from the storage unit 73.

TABLE 1

Table 1 shows the control conditions (learning results) of Q-learning as an example of reinforcement learning. The top row of Table 1 shows the five states st 1-st 5. Specifically, the second to sixth columns show five states st1 to st5, respectively. In addition, the left-most column of Table 1 shows four operations ac 1-ac 4. Specifically, the second to fifth rows show four operations ac 1-ac 4, respectively.

In the example of Table 1, the operation of reducing the output (e.g., voltage) to the heating element 24 by 1% is set to operate ac1 (output change: -1%); the operation of maintaining the output to the heating member 24 is set as operation ac2 (output change: 0%); an operation of increasing the output (e.g., voltage) to the heating member 24 by 1% is set as an operation ac3 (output change: + 1%); the operation of increasing the output (e.g., voltage) to heating element 24 by 1.5% is set as operation ac4 (output change: + 1.5%). The example in table 1 is only a simple example for illustrative purposes, and in practical applications, a larger number of more subdivided operations ac are often set.

The value determined by the combination of state st and operation ac in table 1 is referred to as quality Q (st, ac). For the quality Q, an initial value is given first, and then the quality Q is updated sequentially according to the feedback rw and a known updating formula. The initial value of the quality Q is included in the learning condition shown in fig. 4, for example. The learning condition is input by an operator, for example. For example, an initial value of the quality Q may be stored in the storage unit 73, and a past learning result may be used as the initial value. Further, the learning conditions shown in FIG. 4 include, for example, states st1 to st5 and operations ac1 to ac4 shown in Table 1.

The quality Q is described by taking the state st4 in table 1. In the state st4, the control error is greater than or equal to 0.3 μm and less than or equal to 0.6 μm, indicating that the die lip gap of the target heating bolt 23 is too wide, and therefore it is necessary to increase the output to the heating elements 24 for heating the target heating bolt 23 to increase the degree of thermal expansion of the target heating bolt 23. Therefore, as a result of the learning by the control condition learning unit 72, the quality Q of the operations ac3 and ac4 for increasing the output of the output value heating member 24 is raised, while the quality Q of the operation ac2 for maintaining the output to the heating member 24 and the operation ac1 for decreasing the output to the heating member 24 is lowered.

In the example of table 1, when the control error is, for example, 0.4 μm, the state st is the state st 4. Therefore, the control condition learning unit 72 selects the most appropriate operation ac4 having the highest quality Q in the state st4 and outputs it to the control signal output unit 74.

The control signal output unit 74 increases the control signal ctr to be output to the heating member 24 by 1.5% in accordance with the input operation ac 4. The control signal ctr is, for example, a voltage signal.

Subsequently, when the absolute value of the next control error is smaller than the absolute value of the current control error (0.4 μm), the state observation unit 71 judges that it is appropriate to select the operation ac4 at the current state st4, thereby outputting the feedback rw having a positive value. Accordingly, the control condition learning unit 72 updates the control condition based on the feedback rw, thereby increasing the mass (+5.6) of the operating ac4 in the state st 4. Thus, for the state st4, the control condition learning unit 72 selects the operation ac4 again.

In contrast, when the absolute value of the next control error is larger than the absolute value of the current control error (0.4 μm), the state observation unit 71 determines that it is not appropriate to select the operation ac4 in the current state st4, thereby outputting the feedback rw having a negative value. Accordingly, the control condition learning unit 72 updates the control condition based on the feedback rw, thereby reducing the mass (+5.6) of the operating ac4 in the state st 4. Thus, the quality of operation ac4 in state st4 becomes smaller than the quality (+5.4) of operation ac 3. Therefore, for the state st4, the control condition learning unit 72 selects the operation ac3 instead of the operation ac 4.

The update timing of the control condition is not limited to the subsequent timing, but may be appropriately selected in consideration of the time lag or the like. Furthermore, the learning process can be accelerated by randomly selecting the operation ac at an early stage of the learning process. In addition, although the reinforcement learning is described in Table 1 by taking a simple Q-learning as an example, the present invention is not limited thereto, and any learning algorithm such as Q-learning, Actor-Critic (AC) method, TD-learning, or Monte-Carlo method may be used in the present invention. For example, the learning algorithm may be selected based on the actual situation. For example, when the number of states st and operations AC increases, and there is a case of "combinatorial explosion", the AC method may be employed.

In addition, the AC method often uses a probability distribution function as its policy function. The probability distribution function is not limited to a normal distribution function, and for simplicity, functions such as Sigmoid and Softmax may also be used. The Sigmoid function is the most commonly used function in a neural network, and can also be used since reinforcement learning is a kind of machine learning as in a neural network. The Sigmoid function also has the advantage of being simple and easy to use.

As mentioned above, there are numerous learning algorithms and functions available, and these are appropriately selected.

As described above, since the first embodiment resin film manufacturing apparatus does not employ PID control, it is not necessary to perform parameter adjustment when the process conditions are changed. Further, the control unit 70 updates the control conditions (learning results) based on the feedback rw by reinforcement learning, and selects the most appropriate operation corresponding to the current state st from the updated control conditions. As such, the time and resin material required for adjustment can be reduced as compared with the comparative example even when the process conditions are changed.

< method for producing resin film >

Hereinafter, the method for manufacturing a resin film of the first embodiment is described in detail with reference to fig. 1 and 5. Fig. 5 is a flowchart of a method for controlling a die lip gap in the method for manufacturing a resin film according to the first embodiment.

As shown in fig. 1, in the first embodiment resin film manufacturing method, a film-shaped molten resin 82a is extruded through the gap between the pair of die lips 21a and 22a of the T-die 20.

Subsequently, the resin film 83 as a solidified form of the film-shaped molten resin 82a is conveyed, and the thickness distribution in the width direction of the resin film 83 is measured by the thickness sensor 60.

After that, the control unit 70 performs feedback control of the die lip gap according to the thickness distribution measured by the thickness sensor 60.

Hereinafter, a die lip gap control method in the resin film manufacturing method of the first embodiment is described with reference to fig. 5. The description of fig. 5 will be further described with reference to fig. 4, as needed.

First, as shown in fig. 5, the state observation unit 71 of the control unit 70 shown in fig. 4 calculates a control error of each heating bolt 23 based on the thickness distribution of the resin film 83, and then determines the current state st and the feedback rw for the previously selected operation ac based on the calculated control error (step S1). At the initial moment, since there is no previously (e.g. last) selected operation ac, the feedback rw cannot be determined, but only the current state st.

Subsequently, the control condition learning unit 72 of the control unit 70 first updates the control conditions, which are a combination of the state st and the operation ac, on the basis of the feedback rw, and then selects the most appropriate operation ac corresponding to the current state st from the updated control conditions (step S2).

After that, the control signal output unit 74 of the control unit 70 outputs the control signal ctr to the heating member 24 in accordance with the most appropriate operation ac selected by the control condition learning unit 72 (step S3).

When the manufacturing of the resin film 83 has not been completed (no in step S4), the method returns to step S1, and continues the control. In contrast, when the manufacture of the resin film 83 has been completed (yes in step S4), the method ends. As can be seen, steps S1 to S3 are repeatedly performed until the manufacture of the resin film 83 is completed.

As described above, the first embodiment resin film manufacturing method does not employ PID control, and therefore parameter adjustment is not required when the process conditions are changed. In addition, the method updates the control conditions (learning results) based on the feedback rw by performing reinforcement learning with a computer, and selects the most appropriate operation corresponding to the current state st from the updated control conditions. As such, the time and resin material required for adjustment can be reduced as compared with the comparative example even when the process conditions are changed.

Second embodiment

Hereinafter, a second embodiment resin film manufacturing apparatus will be described. The entire structure of the resin film manufacturing apparatus of the second embodiment is the same as that of the resin film manufacturing apparatus of the first embodiment shown in fig. 1 to 3, and thus, the description thereof is omitted. The second embodiment resin film manufacturing apparatus differs from the first embodiment resin film manufacturing apparatus in the structure of the control unit 70.

Fig. 6 is a block diagram showing the structure of the control unit 70 according to the second embodiment. As shown in fig. 6, the second embodiment control unit 70 includes a state observation unit 71, a control condition learning unit 72, a storage unit 73, and a PID controller 74 a. As can be seen, the second embodiment control unit 70 includes a PID controller 74a that functions as the control signal output unit 74 in the first embodiment control unit 70 shown in fig. 4. The PID controller 74a is a form of a control signal output unit.

Similar to the first embodiment, the state observing unit 71 determines the current state st of each heating bolt 23 and the feedback rw for the previously selected operation ac based on the calculated control error err. Subsequently, the state observing unit 71 outputs the current state st and the feedback rw to the control condition learning unit 72. Further, the state observing unit 71 of the second embodiment also outputs the calculated control error err to the PID controller 74 a.

The control condition learning unit 72 also performs reinforcement learning for each heating bolt 23, in accordance with the first embodiment. Specifically, the control condition learning unit 72 updates the control conditions (learning results) based on the feedback rw, and selects the most appropriate operation ac corresponding to the current state st from the updated control conditions. In the first embodiment, the content of the operation ac selected by the control condition learning unit 72 is a direct change of the output to the heating member 24. In contrast, in the second embodiment, the content of the operation ac selected by the control condition learning unit 72 is a change of the parameter of the PID controller 74 a.

As shown in fig. 6, the parameters of the PID controller 74a are sequentially updated in accordance with the operation ac output by the control condition learning unit 72. The PID controller 74a outputs a control signal ctr to the heating member 24 in accordance with the input control error err. The control signal ctr is, for example, a voltage signal.

Other elements are the same as those of the first embodiment, and thus are not described in detail.

As described above, since the second embodiment resin film manufacturing apparatus employs PID control, parameter adjustment is required when process conditions vary. In the resin film manufacturing apparatus of the second embodiment, the control unit 70 updates the control conditions (learning result) based on the feedback rw by reinforcement learning, and selects the most appropriate operation corresponding to the current state st from the updated control conditions. Wherein the operation ac in reinforcement learning is a change to the PID controller 74a parameter. As such, the time and resin material required for parameter adjustment can be reduced as compared with the comparative example even when the process conditions are changed.

From the above description of the present disclosure, it is apparent that embodiments of the present disclosure may be modified in various ways. Such variations are not to be regarded as a departure from the spirit and scope of the disclosure, and all such modifications as would be obvious to one skilled in the art are intended to be included within the scope of the following claims.