Estimation method, computer program, estimation device, and estimation system

Abstract

Provided are an estimation method and the like capable of estimating the state of a film forming machine. The estimation method acquires detection data on a state of a film forming machine detected by a detector, and estimates an abnormality or life of the film forming machine by inputting the acquired detection data into an estimation model. The estimation model estimates the abnormality or life of the film forming machine when detection data relating to the state of the film forming machine is input.

Description

Technical Field

[0001] This invention relates to estimation methods, computer programs, estimation devices, and estimation systems. Background Technology

[0002] Film forming machines, which solidify molten resin extruded from the outlet of a mold to form a film, are widely used, and technologies for such forming machines have been proposed.

[0003] For example, Patent Document 1 discloses a membrane forming device that can determine whether there is any abnormality in the winding posture of the membrane roll based on the axial change of the radius of the membrane roll. Existing technical documents Patent documents

[0004] Patent Document 1: Japanese Patent Application Publication No. 2019-177508 Summary of the Invention

[0005] However, the technology described in Patent Document 1 concerns the determination of the state of the molded article, not the estimation of the state of the film forming machine. A technology capable of estimating the state of the film forming machine is desired.

[0006] The purpose of this disclosure is to provide a method for estimating the state of a film forming machine, etc.

[0007] One type of estimation method disclosed herein acquires detection data about the state of a membrane forming machine detected by a detector, and estimates the anomaly or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model. The estimation model estimates the anomaly or lifespan of the membrane forming machine in the presence of the detection data about the state of the membrane forming machine.

[0008] One embodiment of the computer program disclosed herein is used to cause a computer to perform the following processing: acquiring detection data about the state of a membrane forming machine detected by a detector, and estimating an anomaly or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model, wherein the estimation model estimates the anomaly or lifespan of the membrane forming machine in the presence of the detection data about the state of the membrane forming machine.

[0009] One type of estimation device disclosed herein includes: an acquisition unit that acquires detection data about the state of a membrane forming machine detected by a detector; and an estimation unit that estimates an anomaly or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model, the estimation model estimating the anomaly or lifespan of the membrane forming machine in the presence of the detection data about the state of the membrane forming machine.

[0010] One embodiment of the estimation system disclosed herein includes a membrane forming machine, a detector, and an estimation device. The estimation device includes: an acquisition unit that acquires detection data about the state of the membrane forming machine detected by the detector; and an estimation unit that estimates an anomaly or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model, wherein the estimation model estimates the anomaly or lifespan of the membrane forming machine in the presence of the detection data about the state of the membrane forming machine. Invention Effects

[0011] Based on this disclosure, the state of the film forming machine can be estimated. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the presumed system of the first embodiment. Figure 2 This is a block diagram illustrating an example of the configuration of a data collection device. Figure 3 This is a block diagram illustrating an example of the configuration of the estimation device. Figure 4 This is a diagram representing an example of the content of information stored in a structured information database (DB). Figure 5 This is an explanatory diagram representing an outline of the hypothetical model. Figure 6 This is a flowchart illustrating an example of the processing sequence performed by the presumed device. Figure 7 This is a schematic diagram illustrating an example of a screen displayed on a terminal device showing an anomaly and an estimated lifespan. Figure 8 This is a flowchart illustrating an example of the processing sequence executed by the estimation device in the second embodiment. Detailed Implementation

[0013] The present invention will be specifically described with reference to the accompanying drawings illustrating embodiments thereof. (First Embodiment) Figure 1 This is a schematic diagram of the estimation system 100 according to the first embodiment. The estimation system 100 includes a film forming machine 1, a plurality of detectors 2, and an estimation device 4 as its main components. The estimation device 4 is communicatively connected to a data collection device 3 and terminal devices 5a and 5b via a network N1 such as the Internet.

[0014] Figure 1 Although a film forming machine 1 and a data collection device 3 are shown in the diagram, multiple data collection devices 3 (not shown) are connected to the presumed device 4 via a network N1. One or more film forming machines 1 are connected to the data collection devices 3. Each data collection device 3 and the one or more film forming machines 1 connected to the data collection device 3 are, for example, located in the same factory.

[0015] The estimation device 4 is a device capable of various information processing, transmission, and reception operations, such as a server computer, personal computer, or quantum computer. The estimation device 4 collects molding information from multiple film forming machines 1 and estimates the malfunctions or lifespan of each film forming machine 1 based on the collected molding information. The molding information includes operational data and detection data, which will be described later. The estimation device 4 sends the estimated malfunctions or lifespan to terminal devices 5a and 5b.

[0016] Terminal devices 5a and 5b are information processing terminals with communication and display functions, such as personal computers, tablets, and smartphones. Terminal device 5a is used by the user of the film forming machine 1. The user of the film forming machine 1 includes legal entities or other organizations that own the film forming machine 1, and operators belonging to that organization. Terminal device 5b is used by service providers such as sales managers and maintenance managers associated with the user's film forming machine 1. There can be two or more terminal devices 5a and 5b.

[0017] <Membrane Forming Machine 1> The film forming machine 1 forms a resin film (membrane) by extrusion molding. The film forming machine 1 includes an extruder 11, a casting device 12, an MD stretching device 13, a TD stretching device 14, a winding machine 15, and a control device 16.

[0018] The extruder 11 is, for example, a single-spindle extruder or a twin-spindle extruder, and includes a raw material supply section 111 for feeding resin raw materials, a cylinder 112, a screw 113, and a die 114. The raw material supply section 111 includes a hopper for storing resin raw materials and a feeder for feeding resin raw materials, discharging the resin raw materials fed into the hopper into the cylinder 112 at a fixed flow rate. The screw 113 is rotatably inserted into a hole in the cylinder 112, and extrudes the resin raw materials discharged from the raw material supply section 111 in the extrusion direction. Figure 1 The material is conveyed (to the right) for melting and mixing. The screw 113 rotates at a specified speed by a drive force output from a drive unit 115, which includes, for example, a motor and a reducer (not shown). The extruder 11 extrudes the molten resin material into a film through a narrow gap (die lip) at the front end of the die 114. The die 114 is, for example, a T-die.

[0019] The casting apparatus 12 includes a plurality of casting rollers 121 for cooling and shaping the hot molten material extruded from the die 114. The casting rollers 121 rotate at a predetermined speed by a driving force output from a drive unit 122 including, for example, a motor and a speed reducer (not shown). The plurality of casting rollers 121 includes a first roller 121a and a second roller 121b.

[0020] The first roller 121a, for example, is a metal roller with a temperature adjustment section (not shown) for cooling the molten material, and is pivotally supported below the die 114. The first roller 121a holds the film-like molten material extruded from the die 114 between itself and the second roller 121b, and together with the second roller 121b, cools the film-like molten material in a short time, simultaneously forming it into a film (sheet). The casting apparatus 12 controls the film thickness within a specified range. Thus, an unstretched film is obtained. The temperature adjustment method for the first roller 121a is not particularly limited, but methods based on heat media such as air, water, or oil, or methods using electric heaters and dielectric heating, are possible. Figure 1 In the example shown, the plurality of casting rolls 121 also include rolls for cooling or conveying the melt.

[0021] The MD stretching device 13 includes a plurality of tension rollers 131, through which the unstretched film conveyed from the casting device 12 is interlaced and stretched in the MD direction. The MD direction is the direction along the film conveying direction, also referred to as the longitudinal direction. The tension rollers 131 rotate at a specified speed by a driving force output from a drive device 132 including, for example, a motor and a reducer (not shown).

[0022] The multiple tension rollers 131 include heating rollers with temperature adjustment sections for heating the film and cooling rollers with temperature adjustment sections for cooling the film. As a method for adjusting the temperature of the tension rollers 131, a method similar to that described for the casting rollers 121 can be used. The film is heated to a predetermined temperature range suitable for stretching while in contact with the heating rollers, and then stretched in the MD direction using the difference in rotational speeds of the cooling rollers. The stretching ratio in the MD direction can be adjusted by the speed ratio of each tension roller 131. It should be noted that... Figure 1 This is just an example; the number of tension rollers 131 is not limited to this. Figure 1 Example shown.

[0023] The TD stretching device 14 stretches the membrane that has been longitudinally stretched by the MD stretching device 13 in the TD direction. The TD direction is the direction that intersects the membrane conveying direction (the width direction of the membrane), also known as the transverse direction.

[0024] The TD stretching device 14 is, for example, a tenter frame stretching device, which has a heating device such as a hot air blowing device (not shown) to heat the film to a specified temperature range that allows it to be stretched, and then performs transverse stretching.

[0025] The TD stretching device 14 includes an oven 141 equipped with heating devices such as a hot air blowing device. Inside the oven 141, a traveling mechanism is provided, including a pair of left and right tracks 142 and a plurality of connecting rods 143 arranged on the tracks 142. The tracks 142 are arranged to extend in the TD direction in a downstream direction toward the MD direction. The plurality of connecting rods 143 form a circulating chain. The circulating chain composed of the plurality of connecting rods 143 travels on the tracks 142 at a predetermined speed by a rotational driving force output from a connecting rod drive device 145 arranged on the outside of the oven 141.

[0026] Clamps 144, serving as membrane holders, are mounted on multiple connecting rods 143. The number of connecting rods 143 and clamps 144 on the left and right tracks 142 are the same. That is, the TD stretching device 14 includes a pair of connecting rods 143 on the left and right sides, and a pair of clamps 144 on the left and right sides. The pair of connecting rods 143 and the pair of clamps 144 are arranged on the tracks 142 in a row along the TD direction, located at the left and right ends of the membrane. By pre-synchronizing the drive device, the pair of clamps 144 are adjusted so that there is no phase difference between the clamps 144, that is, the conveying cycle is consistent. The TD stretching device 14 can be a bearing clamp type or a sliding clamp type.

[0027] At the inlet of the TD stretching device 14, the two ends of the membrane are held by clamps 144. While holding the membrane, the clamps 144, together with the connecting rod 143, are conveyed in the membrane conveying direction, traveling on the track 142, thereby conveying the membrane in the membrane conveying direction. Inside the oven, both sides of the membrane are heated by air blown from hot air blowing devices located above and below the traveling mechanism, and the membrane is stretched in the width direction as the clamps 144 holding the membrane at both ends move. The stretching ratio in the TD direction can be adjusted by the amplitude of the track 142 in the TD direction. The oven 141 is divided into multiple zones, for example, in the conveying direction, and the set temperature or airflow can be adjusted for each zone. These zones include a preheating zone, a stretching zone, and a heat-setting zone.

[0028] The clamp 144 releases the film at the outlet of the TD stretching device 14. The film, stretched in the width direction, is wound by the winder 15 via a traction machine (not shown). It should be noted that the TD stretching device 14 is not limited to a clamp-type tenter frame; for example, it can also be a needle-type tenter frame with needles serving as film holders. The description of the various parts of the TD stretching device 14 described above is just one example; the configuration of temperature and airflow adjustment, the connecting rod 143 of the track 142, etc., is not particularly limited.

[0029] Except for specific time points such as the occurrence of equipment malfunctions and pre-set stop periods, the membrane forming machine 1 usually operates continuously without stopping.

[0030] The control device 16 is a computer that controls the operation of the film forming machine 1. The control device 16 includes a control unit such as a CPU (Central Processing Unit) not shown, a storage unit such as ROM (Read Only Memory), RAM (Random Access Memory), and hard disk, a communication unit that transmits and receives information with external devices via wired or wireless means, a display unit such as a liquid crystal display or an organic EL (Electro Luminescence) display that displays various images to the user, and an operation unit that receives operations from the user.

[0031] The control device 16 is a computer that controls the operation of the film forming machine 1. The control device 16 receives settings for the forming condition parameters of the film forming machine 1 via an operation unit or a communication unit. Additionally, the control device 16 receives detection data detected by the detector 2 from the data collection device 3 via the communication unit. Based on the received forming condition parameters and detection data, the control device 16 controls the operation of the film forming machine 1. Furthermore, the control device 16 sends operation data indicating the operating status of the film forming machine 1 to the data collection device 3 via the communication unit. The data collection device 3 then sends the received operation data to the estimation device 4. It should be noted that the operation data may also be sent directly to the estimation device 4 from devices such as the motor or inverter in the film forming machine 1, without going through the collection device 3.

[0032] For the film forming machine 1, forming condition parameters that determine the forming conditions are set, and the machine operates according to these forming condition parameters. Examples of forming condition parameters include, for example, the feed rate of the extruder 11, the rotational speed of the screw 113, the rotational speed of the motor of the drive unit 115, the die lip opening of the die 114, the rotational speed of the casting roller 121 in the casting device 12, the rotational speed of the motor of the drive unit 122, and the film discharge rate, the rotational speed of the tension roller 131 in the MD stretching device 13, the rotational speed of the motor of the drive unit 132, the stretching ratio, and the film discharge rate, and the linear speed of the film conveying line in the TD stretching device 14, the temperature inside the oven 141 (especially each area), the pressure inside the oven 141 (especially each area), the air volume or velocity at the inlet and outlet of the oven 141 (especially the air blowing nozzles of the hot air blowing device near the inlet and outlet), the stretching angle, the stretching ratio, and the film discharge rate, etc. It should be noted that the membrane speed, membrane width, and membrane thickness can also be set instead of the membrane discharge rate. The optimal forming condition parameters change continuously during operation depending on the environment of the membrane forming machine 1 and the state of the formed product.

[0033] Examples of operating data for the film forming machine 1 include, for example, the feed rate of the extruder 11, the temperature of the cylinder 112, the rotational speed of the screw 113, the rotational speed of the motor, the torque of the motor, and the motor current; the rotational speed of the casting roller 121 in the casting device 12, the rotational speed of the motor, the torque of the motor, and the motor current; the rotational speed of the tension roller 131 in the MD stretching device 13, the rotational speed of the motor, the torque of the motor, and the motor current; and the linear speed and linear tension in the TD stretching device 14. These operating data correspond to the setpoints for each item, that is, the control values ​​based on the control device 16.

[0034] <Detector 2> Detector 2 is a sensor that detects physical quantities representing the state of membrane forming machine 1 in a time-series manner. The state of membrane forming machine 1 detected by detector 2 includes not only the state of membrane forming machine 1 itself, but also the state of the membrane (resin) formed by membrane forming machine 1. Detector 2 is connected to data collection device 3 via wired or wireless means, and outputs the detection data obtained through detection directly or indirectly to data collection device 3.

[0035] The estimation system 100 has multiple detectors 2 that detect various types of detection data. A portion of the multiple detectors 2 is connected to a data collection device 3, from which the data collection device 3 acquires detection data. A portion of the multiple detectors 2 is connected to a control device 16, through which the data collection device 3 acquires detection data from the detectors 2. It should be noted that... Figure 1 The diagram omits part of the connection line between detector 2 and data collection device 3 or control device 16.

[0036] Detector 2 includes detectors for detecting sensor values ​​required for the operation control of film forming machine 1, and detectors for detecting sensor values ​​required for implementing anomaly estimation or lifespan estimation of film forming machine 1 based on this estimation method. Detector 2 can be a separate component from film forming machine 1, or it can be a component incorporated into film forming machine 1.

[0037] As detector 2, examples include accelerometers, velocity sensors, displacement sensors, AE (Acoustic Emission) sensors, viscometers, gel counters, strain sensors, crystal orientation sensors, anemometers, flow meters, thermometers, surface potential sensors, laser sensors, infrared sensors, X-ray sensors, position sensors, tension sensors, force gauges, torque sensors, encoders, current sensors, gyrometers, angle sensors, pressure gauges, weight gauges, length sensors, light sensors (cameras), flow meters, voltmeters, timers, and power meters.

[0038] The detection data detected by detector 2 can include, for example, acceleration, velocity, displacement, viscosity, gel quantity, strain, crystal orientation, wind speed, air volume, temperature, charge, position, phase difference, tension, torque, rotational speed, angle, pressure, weight, length, thickness, image, flow rate, current, voltage, time, force, and power consumption. In this specification, the detection data detected by detector 2 is not limited to sensor values ​​directly detected by detector 2, but also includes calculated values ​​indirectly derived from those sensor values.

[0039] As for the detection data of the extruder 11 and the detector 2 for that detection data, examples include an accelerometer that detects the acceleration (vibration) of the reducer, a gyrometer that detects the rotational speed of the screw 113 or the motor, a gyrometer that detects the rotational speed of the screw 113 or the motor, a torque sensor that detects the torque of the motor, a current sensor that detects the current of the motor, a viscometer that detects the viscosity of the resin (especially the viscosity of the resin after discharge from the mold 114), a contact thermometer or a non-contact thermal imaging camera that detects the temperature inside the cylinder 112 or the resin (especially the resin after discharge from the mold 114), a laser displacement sensor that detects the die lip opening of the mold 114, and a weight gauge that detects the amount of feed supplied by the feeder.

[0040] Examples of detectors 2 that can be used to detect data about the casting apparatus 12 and the data include an acceleration sensor that detects the acceleration (vibration) of the bearing of the casting roller 121, a speed sensor that detects the rotational speed of the casting roller 121 or the motor, a gyroscope that detects the rotational speed of the casting roller 121 or the motor, a torque sensor that detects the torque of the motor, a current sensor that detects the current of the motor, a thermometer or thermal imaging camera that detects the temperature of the casting roller 121 or the film, a laser sensor that detects the width of the film, a laser thickness sensor that detects the thickness of the film, and a gel counter that detects the amount of gel in the film (especially the film downstream of the casting roller 121).

[0041] As the detector 2 for the detection data of the MD stretching device 13, examples include an acceleration sensor that detects the acceleration (vibration) of the bearing of the tension roller 131, a gyroscope that detects the rotational speed of the tension roller 131 or the motor, a speed sensor that detects the rotational speed of the tension roller 131 or the motor, a torque sensor that detects the torque of the motor, a current sensor that detects the current of the motor, a thermometer or thermal imaging camera that detects the temperature of the tension roller 131 or the film, a laser sensor that detects the width of the film, and a laser sensor that detects the thickness of the film.

[0042] As for the detection data of the TD stretching device 14 and the detector 2 for this detection data, examples include: an acceleration sensor that detects the acceleration (vibration) of the TD stretching device 14 (especially the track 142 near the start and end of the stretching zone in the oven 141); a velocity sensor that detects the velocity (vibration); a displacement sensor that detects the displacement (vibration); an AE sensor that detects elastic waves; a strain sensor that detects the strain of the clamp 144; a laser sensor that detects the phase difference (difference in the passing time of the reference position) of the left and right pair of clamps 144; a light sensor or photoelectric sensor; a thermometer or thermal imaging camera that detects the temperature inside the oven 141 (especially each area); a pressure gauge that detects the pressure inside the oven 141 (especially each area); and a detector that detects the inlet and outlet of the oven 141 (especially the hot air near the inlet and outlet). An anemometer for measuring the airflow in the air blowing nozzles of the blowing device, an anemometer for measuring the air velocity at the inlet and outlet of the oven 141 (especially at the air blowing nozzles of the hot air blowing device near the inlet and outlet), a pressure gauge for measuring the pressure at the inlet and outlet of the oven 141 (especially at the air blowing nozzles of the hot air blowing device near the inlet and outlet), a rotatometer for measuring the fan speed in the hot air blowing device, a speed sensor for measuring the operating line speed of the membrane conveying line, a tension sensor for measuring the tension of the membrane conveying line or clamp 144, a crystal orientation sensor for measuring the degree of crystal orientation of the membrane (especially the membrane at the downstream end of the TD stretching device 14), a gel counter for measuring the amount of gel in the membrane (especially the membrane at the downstream end of the TD stretching device 14), a laser sensor for measuring the width of the membrane, and a laser sensor for measuring the thickness of the membrane, etc.

[0043] As a detector 2 for the detection data of the winding machine 15 and the detection data, for example, a surface potential sensor capable of detecting the charge of the film is used.

[0044] Detector 2 is located at a suitable location in or around membrane forming machine 1 to detect components at desired locations within membrane forming machine 1 and the state of the membrane. Detector 2 is preferably characterized by heat resistance, water resistance, oil resistance, etc., depending on its location. It should be noted that detector 2 and the content of the detection data are not limited to the examples described above. Furthermore, the detection data described is not limited to the data detected by the illustrated detector 2, and detector 2 is not limited to detectors used to detect the illustrated detection data.

[0045] <Data Collection Device 3> Figure 2This is a block diagram illustrating an example configuration of the data collection device 3. The data collection device 3 is a computer, including a control unit 31, a storage unit 32, a communication unit 33, and a data input unit 34. The storage unit 32, the communication unit 33, and the data input unit 34 are connected to the control unit 31. The data collection device 3 may be, for example, a PLC (Programmable Logic Controller).

[0046] The control unit 31 includes arithmetic processing circuits such as a CPU (Central Processing Unit), a multi-core CPU, an ASIC (Application Specific Integrated Circuit), and a FPGA (Field-Programmable Gate Array), internal storage devices such as ROM (Read Only Memory) and RAM (Random Access Memory), and I / O terminals. The control unit 31 executes the control program stored in the storage unit 32 (described later), collects forming information including detection data and operational data, and performs processing to send it to the estimation device 4. It should be noted that the functional units of the data collection device 3 can be implemented in software, in hardware, or in a combination thereof.

[0047] Storage unit 32 includes non-volatile storage devices such as hard disk, EEPROM (Electrically Erasable Programmable ROM), and flash memory. Storage unit 32 stores control programs used to enable the computer to perform the collection and processing of formed information.

[0048] The communication unit 33 includes communication equipment that enables communication via networks N1 and N2. The communication unit 33 is connected to the control device 16 via network N2, such as a LAN (Local Area Network). The control unit 31 can send and receive various information between the control devices 16 via the communication unit 33. Furthermore, the communication unit 33 is connected to the estimation device 4, which is the cloud of network N1, via a relay device (not shown). The relay device may be, for example, a router or gateway. The control unit 31 can send and receive various information between the estimation device 4 via the relay device and the communication unit 33.

[0049] The data input unit 34 is an input interface for receiving signals output from the detector 2. The detector 2 is connected to the data input unit 34. The control unit 31 acquires detection data output from the detector 2 at any time via the data input unit 34.

[0050] The data collection device 3 and the control device 16 are not limited to separate devices, but can also be composed of a common single device.

[0051] <Presumption Device 4> Figure 3 This is a block diagram illustrating an example configuration of the estimation device 4. The estimation device 4 includes a control unit 41, a storage unit 42, and a communication unit 43. The storage unit 42 and the communication unit 43 are connected to the control unit 41. The estimation device 4 can be configured by multiple computers performing distributed processing, or it can be implemented using multiple virtual devices located within a server, or it can be implemented using a cloud server.

[0052] The control unit 41 includes a CPU, multi-core CPU, ASIC, FPGA and other arithmetic processing circuits, internal storage devices such as ROM and RAM, I / O terminals, etc. The functional units of the estimation device 4 can be implemented through software, hardware, or a combination thereof.

[0053] Storage unit 42 includes, for example, non-volatile storage such as hard disk, flash memory, or SSD (Solid State Drive). Storage unit 42 may also be an external storage device connected to estimation device 4. Storage unit 42 stores various computer programs and data for reference by control unit 41. In this embodiment, storage unit 42 stores a program 4P for causing the computer to perform processing related to the estimation of abnormalities or lifespan of film forming machine 1, and a molding information DB (Database) 421 and estimation model 422 required for the execution of program 4P. Molding information DB 421 is a database storing molding information received from each data collection device 3. Estimation model 422 is a learning model generated through machine learning. It is assumed that estimation model 422 utilizes program modules that constitute artificial intelligence software.

[0054] The computer program (computer program product) including program 4P can be provided by a non-transitory recording medium 4A that readablely records the computer program. Storage unit 42 stores the computer program read from the recording medium 4A by a reading device (not shown). Recording medium 4A may be, for example, a magnetic disk, optical disk, semiconductor memory, etc. Alternatively, the computer program may be downloaded from an external server connected to a communication network and stored in storage unit 42. Program 4P can be a single computer program or a combination of multiple computer programs; furthermore, it can be executed on a single computer or on multiple computers interconnected via a communication network.

[0055] The communication unit 43 includes a communication device that enables communication via network N1. The control unit 41 is capable of sending and receiving various information between the data collection device 3 and the terminal devices 5a and 5b via the communication unit 43.

[0056] <Molding Information DB421> Figure 4This diagram illustrates an example of the information stored in the molding information DB421. The molding information DB421 stores records that associate information such as equipment ID, date and time, operation data, and test data, using data ID as a key. The equipment ID represents the equipment identification code of the film forming machine 1. The date and time represent the year, month, day, and time when the operation data or test data was obtained and stored as a record. The operation data includes the various types of operation data mentioned above. The test data includes the various types of test data mentioned above. Whenever the estimation device 4 receives operation data and test data from the data collection device 3, it stores them in the molding information DB421 in chronological order. The content of the molding information DB421 is updated continuously.

[0057] <Presumed Model 422> Figure 5 This is an explanatory diagram showing the outline of the presumed model 422. Figure 5 The following diagram illustrates a estimation model 422, which consists of a first estimation model 422a and a second estimation model 422b, as an example. The first estimation model 422a is a learning model that takes molding information, including detection data and operating data of the film forming machine 1, as input and outputs abnormality data indicating whether or not the film forming machine 1 has any abnormalities. The second estimation model 422b is a learning model that takes molding information, including detection data and operating data of the film forming machine 1, as input and outputs failure probability data indicating the failure probability of the film forming machine 1 at a specified time point. Since the first estimation model 422a and the second estimation model 422b have the same configuration, the following explanation will focus on the first estimation model 422a.

[0058] The first estimation model 422a is, for example, a neural network. The first estimation model 422a includes an input layer that receives the shaping information, an output layer that outputs data indicating the presence or absence of abnormal data or the probability of failure, and intermediate layers (hidden layers) that extract features. The intermediate layers may also contain convolutional layers, pooling layers, and fully connected layers. The intermediate layers have multiple nodes that extract features from the input data and transmit the features extracted using various parameters to the output layer. With the shaping information input to the input layer, calculations are performed in the intermediate layers based on the learned parameters, and the output layer outputs information representing the presence or absence of abnormal data or the probability of failure.

[0059] For the input layer of the first estimation model 422a, the various types of detection data and operational data mentioned above are input. The detection data includes detection data detected by detector 2 for state estimation. The detection data may also include detection data detected by detector 2 for operational control.

[0060] In this embodiment, various feature quantities derived from detection data and operation data are input as input information to the first estimation model 422a. The detection data and operation data used for feature quantity calculation include the various detection data and operation data described above. Alternatively, the detection data and operation data input to the first estimation model 422a may be combinations of appropriate detection data and operation data selected from the example detection data and operation data described above. The detection data and operation data are, for example, time-series data of a portion of the entire period from the start of operation to the data acquisition time point, or a predetermined number of detections. The amount of detection data and operation data can be one cycle of the film forming machine 1. By using data sets from multiple time points, it is expected that noise will be reduced and estimation accuracy improved.

[0061] It should be noted that by setting the feature quantities of the detection data and operation data as inputs to the first estimation model 422a, the characteristics of each data can be reflected and the estimation accuracy of the model is expected to be improved. However, the detection data and operation data themselves can also be used as inputs to the first estimation model 422a.

[0062] As feature quantities obtained from detection data and operational data, these can include, for example, feature quantities that represent the characteristics of time series such as the root mean square (RMS) of various data, the root mean square (RMS) of the time-synchronized average of various data, and the energy ratio of the time-synchronized average of various data; feature quantities that represent the characteristics of frequency components such as amplitude, intensity, frequency, and image data obtained through frequency analysis such as Fourier transform, fast Fourier transform, and Hilbert spectrum analysis of various data; and values ​​obtained based on these feature quantities such as time derivatives and differences of the aforementioned feature quantities. Different feature quantities can be calculated for each type of detection data and operational data. For example, the estimation device 4 stores a table (not shown) in advance in the storage unit 42, in which a correspondence is established between the formula used for calculating the feature quantity and the data items used for calculating the feature quantity for each feature quantity.

[0063] The output layer of the first presumption model 422a has nodes that output data indicating the presence or absence of anomalies. It should be noted that the presumption of anomalies is not limited to being represented by the presence or absence of anomalies. For example, the presumption model 422 can be configured to output an anomaly degree that numerically represents the degree of anomaly, or it can be configured to output an anomaly level that represents the degree of anomaly in multiple grades.

[0064] The second estimation model 422b has the same structure as the first estimation model 422a, and uses various feature quantities derived from the detection data and operational data as inputs. It should be noted that the feature quantities input to the first estimation model 422a can be the same as or different from those input to the second estimation model 422b. Furthermore, the detection data and operational data used to calculate the feature quantities input to the first estimation model 422a can be the same as or different from those input to the second estimation model 422b. The second estimation model 422b can also be configured to further use anomaly presence / absence data estimated by the first estimation model 422a as input elements.

[0065] The output layer of the second estimation model 422b has nodes that output failure probability data. The failure probability data represents the probability of failure of the membrane forming machine 1 at a future point in time after a specified period from now, for example, the probability that the membrane forming machine 1 will fail in three months. The failure probability represents the lifespan of the membrane forming machine 1. It should be noted that the estimation model 422 can also output a value for the remaining lifespan of the membrane forming machine 1 (how long afterward the membrane forming machine 1 will fail) as a lifespan estimate.

[0066] Training data, labeled with data indicating the presence or absence of anomalies, is prepared relative to the feature quantities derived from detection data and operational data. This training data is used to enable an unlearned neural network to perform machine learning, thereby generating a first estimation model 422a. The estimation device 4 acquires training data comprising a set of data that establishes a correspondence between past detection data and operational data from the film forming machine 1 and identification results that have identified the presence or absence of actual anomalies. Preferably, the estimation device 4 collects a large amount of detection data and operational data from multiple film forming machines 1 as training data.

[0067] The estimation device 4 inputs the features contained in the training data into the input layer of the first estimation model 422a. After processing in the intermediate layer, it obtains the identification result of the presence or absence of anomalies output from the output layer. The estimation device 4 compares the identification result of the presence or absence of anomalies output from the output layer with the identification result contained in the training data, and optimizes parameters such as the weights (coupling coefficients) between neurons by using error backpropagation to make the identification result output from the output layer closer to the correct value. In the stage before learning begins, initial settings are assigned to the definition information describing the first estimation model 422a. When learning is completed by meeting the prescribed benchmarks through error, number of learning iterations, etc., the optimized parameters are obtained. If learning is complete, the estimation device 4 stores the definition information of the first estimation model 422a as the learned first estimation model 422a in the storage unit 42.

[0068] Similarly, a second presupposition model 422b can be generated. Through the above processing, a first presupposition model 422a, which is learned in a way that can appropriately identify the presence or absence of anomalies relative to the feature quantities of the formed information, and a second presupposition model 422b, which is learned in a way that can appropriately identify the probability of failure relative to the feature quantities of the formed information, are constructed. It should be noted that the composition of the first presupposition model 422a and the second presupposition model 422b is not limited to the same form; they can also be different. For example, the first presupposition model 422a can be a binary classification model, and the second presupposition model 422b can be a regression model that outputs continuous values.

[0069] The estimated model 422 is not limited to a model generated and learned by the estimated device 4. The estimated model 422 may also be a model that has been learned and sent to the estimated device 4 by an external server and stored in the storage unit 42. The estimated model 422 may also be generated by an external server and learned by the estimated device 4.

[0070] The composition of the first estimation model 422a and the second estimation model 422b is not limited, as long as they can estimate anomalies or lifespans relative to the detection data and operational data. The first estimation model 422a and the second estimation model 422b can be, for example, a Recurrent Neural Network (RNN), a Convolutional Neural Network (CNN), a Graph Neural Network (GNN), a Transformer, a Support Vector Machine, a Decision Tree, eXtreme Gradient Boosting (XGBoost), Logistic Regression, Random Forest, or other models based on learning algorithms. The first estimation model 422a and the second estimation model 422b can also be models learned through unsupervised learning. The first estimation model 422a and the second estimation model 422b include multiple models and can be constructed using ensemble learning methods such as guided bagging, boosting, or stacking. The first presumption model 422a and the second presumption model 422b can presume anomalies or lifetimes using a rule-based approach.

[0071] In this embodiment, although the first estimation model 422a for estimating the presence or absence of an anomaly and the second estimation model 422b for estimating the probability of failure are configured separately, they can also be configured to use a single learning model to estimate the presence or absence of an anomaly and the probability of failure. Alternatively, the estimation device 4 may have only either the first estimation model 422a or the second estimation model 422b, and only estimate either the presence or absence of an anomaly or the probability of failure.

[0072] The estimation model 422, by using a variety of detection and operational data as input elements, can accurately estimate anomalies or lifespans considering diverse characteristics. For example, it is believed that inputting mechanical vibration, represented by acceleration, can account for the deterioration state of device components. It is also believed that inputting resin viscosity can account for changes in resin properties. Furthermore, it is believed that inputting gel quantity can account for quality anomalies caused by the presence or absence of unmixed portions and foreign matter in the resin raw material. It is also believed that inputting clamp strain can account for the deterioration state of the device. Furthermore, it is believed that inputting crystal orientation degree can account for quality anomalies. It is also believed that inputting wind speed, airflow, or pressure in oven 141 can account for the environment within oven 141. Furthermore, it is believed that inputting the temperature of each detection point and the object being detected can account for diverse temperature environments. Furthermore, it is believed that inputting membrane charge can account for poor membrane winding and quality anomalies around the winding machine 15. Finally, it is believed that inputting synchronization error, represented by the phase difference between the left and right clamps, can account for sliding load and quality changes caused by device load and deterioration. It is believed that by inputting the line tension, the wear condition of the track can be taken into account. It is also believed that by inputting the motor torque, speed, and rotational speed of each detection part and the object being detected, the deterioration condition of the drive unit can be taken into account.

[0073] Figure 6 This is a flowchart illustrating an example of the processing sequence executed by the estimation device 4. The control unit 41 of the estimation device 4 executes the following processes according to the program 4P stored in the storage unit 42. The control unit 41 initiates the following processes at predetermined or appropriate intervals.

[0074] The control unit 41 of the estimation device 4, acting as an acquisition unit, acquires molding information, including detection data and operation data during continuous operation, via the data collection device 3, and stores the acquired molding information in the molding information DB421 (step S10). A correspondence is established between the device ID of the film forming machine 1 and the molding information. The control unit 41 can acquire multiple detection data and operation data simultaneously, or it can acquire them individually at the time points when each detection data and operation data is detected. The time point for receiving the molding information can be a relatively quiet time period in the network N2.

[0075] The control unit 41 determines whether to perform the estimation process (step S11). If the estimated process is not performed because the preset start conditions are not met (S11: No), the control unit 41 returns the process to step S11 and waits until the start conditions are met. The start conditions may include the acquisition of a preset number of detection data and operation data, and a preset estimated time point, etc.

[0076] If the pre-set start conditions are met and the pre-set process is determined to be executed (S11: Yes), the control unit 41 calculates various characteristic quantities based on the acquired time-series detection data and operation data (step S12). The control unit 41 refers to, for example, a table that establishes a correspondence between formulas used for characteristic quantity calculation and data items, and determines the calculation method, detection data, and operation data used for each characteristic quantity calculation.

[0077] According to its function as an estimation unit, the control unit 41 inputs the calculated detection data and characteristic quantities of the operation data into the first estimation model 422a (step S13), and obtains the abnormality data output from the first estimation model 422a (step S14). Through the processing of steps S13 and S14, the abnormality of the film forming machine 1 is estimated.

[0078] According to its function as an estimation unit, the control unit 41 inputs the calculated detection data and characteristic quantities of the operation data into the second estimation model 422b (step S15), and obtains the failure probability data output from the second estimation model 422b (step S16). Through the processing of steps S15 and S16, the lifespan of the film forming machine 1 is estimated.

[0079] The control unit 41 determines whether to output an alarm (step S17). For example, if the estimated result is abnormal and the failure probability meets or exceeds at least one of the preset thresholds, the control unit 41 determines to output an alarm. If the estimated result is not abnormal and the failure probability is less than the preset threshold, the control unit 41 determines not to output an alarm.

[0080] If it is determined that no alarm will be output (S17: No), the control unit 41 terminates the process. If it is determined that an alarm will be output (S17: Yes), the control unit 41 generates a screen indicating the abnormality of the film forming machine 1 and the estimated lifespan (step S18). The screen indicating the estimated result is an example of reporting alarm information about the abnormality and lifespan of the film forming machine 1. The control unit 41 sends the screen indicating the generated estimated result to the terminal device 5a of the user corresponding to the device ID of the film forming machine 1 and the terminal device 5b of the service provider (step S19). The control unit 41 may also send the estimated result to the control device 16. The control unit 41 terminates the series of processes.

[0081] The control unit 41 repeatedly performs the above-described processing during the continuous operation of the film forming machine 1. Based on the detection data and operation data detected during continuous operation, it performs anomaly and lifespan estimation while continuous operation continues, thereby feeding back the estimation results to the operation of the film forming machine 1.

[0082] While the above illustrates an example of the presumption device 4 performing a series of processes, the processing entity for each process is not limited. The processes in the flowchart can also be performed by, for example, the control device 16, the data collection device 3, or the terminal device 5. Alternatively, the presumption model 422 can be deployed on the control device 16 or the data collection device 3, with the control device 16 or the data collection device 3 performing the presumption processes for anomalies and lifespan.

[0083] In the above process, the anomaly estimation processing in steps S13 and S14, and the lifetime estimation processing in steps S15 and S16, can be performed in a different order or in parallel. Either the anomaly estimation processing or the lifetime estimation processing can also be omitted.

[0084] The estimation device 4 can also be used to relearn the estimation model 422 based on molding information obtained during continuous operation. For example, the estimation device 4 receives identification results of anomalies or lifespans relative to the obtained molding information within a specified period from the service provider. The estimation device 4 relearns the estimation model 422 relative to the molding information, using information labeled with the identification results of the acquired anomalies or lifespans as training data. The training data used for relearning may also include molding information obtained from multiple film forming machines 1.

[0085] The estimation device 4 can evaluate the accuracy of the estimation model 422, for example, through evaluation methods such as cross-validation, and update the estimation model 422 if the evaluation indicators meet the specified benchmarks. The estimation device 4 can also receive confirmation from the service provider that the estimation model 422 needs to be updated. Thus, the accuracy of anomaly or lifespan estimations can be improved through the application of this system.

[0086] Figure 7 This is a schematic diagram of an example of a screen 70 that displays an anomaly and an estimated lifespan result shown in terminal devices 5a and 5b. The screen 70 has a result display unit 701 that displays the estimated result of whether there is an anomaly and the probability of failure.

[0087] The estimation device 4 acquires the estimation results obtained based on the estimation model 422, and displays the acquired abnormality and failure probability on the result display unit 701. The estimation device 4 also outputs alarm information through the screen 70 by displaying the alarm display unit 702 on the screen 70. The alarm display unit 702 displays text indicating abnormality or that the estimated lifespan result is abnormal.

[0088] The screen 70 also includes a trend display unit 703 that displays trends in characteristic quantities. The trend display unit 703 includes a receiving unit 704 that receives the period for displaying the trend and the type of characteristic quantity; and a trend graph display unit 705 that displays the trend graph. The receiving unit 704 has, for example, a drop-down menu. Users or service providers can select any period and characteristic quantity type by opening the drop-down menu. The list of characteristic quantity types in the drop-down contains each characteristic quantity used in the input information of the estimation model 422.

[0089] When the estimation device 4 receives the selection of the period and the type of feature quantity through the terminal devices 5a and 5b, it reads the detection data and operation data for the selected period from the molding information DB421. Based on the read detection data and operation data, the estimation device 4 generates a time series graph representing the trend of the selected feature quantity and displays it in the trend display unit 703. It should be noted that the trend display unit 703 can display the trend of the detection data and operation data instead of or based on the feature quantity.

[0090] The above describes an example of outputting alarm information along with the estimation result, but the estimation device 4 may also output alarm information reporting an anomaly in the film forming machine 1 to the terminal devices 5a and 5b. The alarm information may be in the form of, for example, a message, voice, or alarm tone.

[0091] The estimation device 4 can provide information based on the requirements from the terminal devices 5a and 5b, regardless of whether there are any abnormalities. Figure 7 The screen shown is 70. Each user or service provider representative can view and confirm the status of the film forming machine 1 owned by the user at any time by accessing the service provider's website, for example, using a browser.

[0092] According to this embodiment, the condition of the membrane forming machine 1, such as abnormalities or lifespan, can be accurately estimated based on diverse detection data obtained from the detector 2 that detects the condition of the membrane forming machine 1. The estimated result allows for a correct understanding of the condition of the membrane forming machine 1. In conventional membrane forming machines, the detection data from the detector 2 is primarily used for operational control, and the condition of the membrane forming machine 1 during operation is not considered in estimating abnormalities or lifespan. In this embodiment, a detector 2 is provided for the purpose of maintaining the membrane forming machine 1, thereby enabling the collection of a larger amount of detection data that was previously unavailable in conventional membrane forming machines, and its application to condition analysis.

[0093] By collecting monitoring data during the continuous operation of the membrane forming machine 1 and performing anomaly or lifespan estimations based on the collected data, the status of the membrane forming machine 1 can be quickly assessed and addressed smoothly. Because it can be addressed before the condition deteriorates, equipment maintenance becomes easier, and production downtime can be reduced.

[0094] By using the estimation model 422, anomalies or lifespan of the film forming machine 1 can be easily and accurately estimated. The estimation accuracy of anomalies or lifespan can be improved by using multiple detection data representing the state of the film forming machine 1 as input information to the estimation model 422. Preprocessing for calculating characteristic quantities is performed relative to the detection data and operating data, thereby assigning information suitable for anomaly estimation and lifespan estimation to the estimation model 422, thus improving estimation accuracy.

[0095] The estimation device 4 can accumulate a large amount of data in the cloud and provide the terminal device 5 with the trends of the accumulated data and the estimation results of anomalies or lifespan based on the data. Users and service providers can visually identify the status of the film forming machine 1 at any point in time.

[0096] (Second Implementation) In the second embodiment, the configuration using the estimated model 422 or characteristic quantities corresponding to the film forming machine 1 will be described. The following embodiments mainly describe the differences from the first embodiment. For configurations common to the first embodiment, the same reference numerals will be used and detailed descriptions will be omitted.

[0097] The estimation device 4 of the second embodiment stores various types of estimation models 422 generated by different algorithms in the storage unit 42. Before estimating anomalies or lifespans corresponding to molding information, the estimation device 4 determines the estimation model 422 to be used in estimating anomalies or lifespans based on the type of film forming machine 1 owned by the user and the molding information collected from that film forming machine 1. The determination of the estimation model 422 can be based on the estimation accuracy of each estimation model 422. For example, the accuracy of each estimation model 422 is evaluated by evaluation methods such as cross-validation based on the estimation results of each estimation model 422 relative to the molding information in the film forming machine 1. From all the prepared estimation models 422, the estimation model 422 with the highest estimation accuracy is selected, thereby determining the estimation model 422.

[0098] The estimation device 4 can select multiple estimation models 422 with high estimation accuracy, and use the selected multiple estimation models 422 to estimate anomalies or lifetimes. For example, the estimation device 4 can combine the estimation results obtained based on multiple types of estimation models 422, and estimate anomalies or lifetimes by majority vote or weighted integration.

[0099] The estimation device 4 can also be used to fine-tune the pre-generated estimation model 422 that has been learned, and generate an estimation model 422 specifically for estimating the abnormality or lifespan of the membrane forming machine 1 owned by the user.

[0100] The estimation device 4 can also select, based on the type of film forming machine 1 owned by the user, a set of pre-defined features for use in the input information to the estimation model 422. The features used for input information can be determined, for example, based on their contribution. For instance, using a set of pre-defined features, estimations based on the estimation model 422 can be performed, employing methods such as Shapley Additive Explanation (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), and feature importance in decision tree algorithms to calculate the contribution of the input data to each feature. The estimation device 4 can then select, for example, features with a calculated contribution value greater than a specified value, or a specified number of features selected in descending order of contribution, thereby determining the features used for input information.

[0101] The estimation device 4 establishes a correspondence between the device ID of the determined estimation model 422, the type of the feature quantity used for input information, and the device ID of the film forming machine 1 to which the estimation model 422 and the feature quantity are applicable, and stores it in the storage unit 42.

[0102] Figure 8 This is a flowchart illustrating an example of the processing sequence executed by the estimation device 4 in the second embodiment.

[0103] The control unit 41 of the estimation device 4 establishes a correspondence between the detection data and the molding information of the operation data during continuous operation and the device ID of the film forming machine 1, and stores the obtained molding information in the molding information DB421 (step S21).

[0104] The control unit 41 determines whether to perform the presumption process (step S22). If the presumption process is not performed because the preset start conditions are not met (S22: No), the control unit 41 returns the process to step S22 and waits until the start conditions are met.

[0105] If the pre-set start conditions are met and the pre-set process is deemed to be in the pre-set process (S22: Yes), the control unit 41 selects a pre-set model 422 that corresponds to the acquired device ID of the film forming machine 1 (step S23). In step S23, the control unit 41 determines the pre-set model 422 corresponding to the acquired device ID based on the pre-stored correspondence between the film forming machine 1 and the applicable pre-set model 422.

[0106] Based on the acquired time-series detection data and operation data, the control unit 41 calculates various characteristic quantities (step S24). When the characteristic quantities used are set for each type of the film forming machine 1, the control unit 41 determines the characteristic quantities that correspond to the acquired device ID of the film forming machine 1 based on the pre-stored correspondence between the film forming machine 1 and the applicable characteristic quantities, and calculates the determined characteristic quantities.

[0107] Hereinafter, the control unit 41 can use the selected estimation model 422 and characteristic quantities to estimate anomalies and lifetimes by performing the same process as in the first embodiment.

[0108] According to this embodiment, by using an estimation model 422 and characteristic quantities corresponding to the type of membrane forming machine 1 owned by the user, the estimation accuracy of anomalies or lifespan can be improved. The configuration of the membrane forming machine 1 varies greatly from device to device. By setting the type of estimation model 422 and the characteristic quantities used according to the type of membrane forming machine 1, anomalies or lifespan can be estimated with high accuracy corresponding to the state of the membrane forming machine 1.

[0109] The embodiments disclosed herein are illustrative in all respects and should not be considered limiting. The technical features described in the various embodiments can be combined with each other, and the scope of this invention is intended to include all modifications and equivalent scopes within the scope of the technical solutions. The order shown in each embodiment is not limited. Within the scope of non-contradiction, the order of each process can be changed and executed. In addition, multiple processes can be executed in parallel. The processing entity of each process is not limited. Within the scope of non-contradiction, the processing of each device can also be performed by other devices.

[0110] The items described in each embodiment can be combined with each other. Furthermore, the independent and dependent items described in the technical solution, regardless of their reference form, can be combined with each other in all combinations. Moreover, although the technical solution uses the form of referencing two or more other items (multiple references to multiple items), it is not limited to this. It can also be described using the form of multiple dependent items referencing at least one item (multiple references to multiple items).

[0111] Regarding the above implementation methods, the following notes are further disclosed. (Note 1) One method of estimation, wherein, Acquire detection data regarding the status of the membrane forming machine detected by the detector. The abnormality or lifespan of the membrane forming machine is estimated by inputting the acquired detection data into the estimation model. The estimation model estimates the abnormality or lifespan of the membrane forming machine in the presence of detection data about the state of the membrane forming machine. (Note 2) In the estimation method described in Appendix 1, The detection data is acquired during the continuous operation of the film forming machine. Under continuous operation, the abnormality or lifespan of the film forming machine is inferred based on the detection data. (Note 3) In the estimation method described in Note 1 or Note 2 Derive the feature values ​​derived from the detection data. The abnormality or lifespan of the membrane forming machine is estimated by inputting the derived feature quantities into the estimation model, which estimates the abnormality or lifespan of the membrane forming machine in the presence of feature quantities derived from detection data about the state of the membrane forming machine. (Note 4) In any of the presumption methods described in Appendix 1 to Appendix 3 Information indicating an estimated malfunction or lifespan of the membrane forming machine is sent to the terminal device. (Note 5) In any of the presumption methods described in Appendix 1 to Appendix 4, Based on the time-series detection data, information representing the trend of the detection data is generated. The generated information representing the trend of the detection data is sent to the terminal device. (Note 6) In any of the presumption methods described in Appendix 1 to Appendix 5 The film forming machine includes an extruder, a casting device, an MD stretching device, a TD stretching device, and a winding machine. The detection data includes the physical quantities of the extruder, the physical quantities of the casting device, the physical quantities of the MD stretching device, the physical quantities of the TD stretching device, or the physical quantities of the winding machine. (Note 7) In any of the presumption methods described in Appendix 1 to Appendix 6 The film forming machine includes an extruder, a casting device, an MD stretching device, a TD stretching device, and a winding machine. The detection data includes at least one of the following: acceleration in the extruder, resin viscosity in the extruder, acceleration in the casting device, amount of gel in the film in the casting device, acceleration in the MD stretching device, acceleration in the TD stretching device, wind speed in the TD stretching device, air volume in the TD stretching device, strain of the film holder in the TD stretching device, phase difference between the left and right pairs of film holders in the TD stretching device, crystal orientation in the TD stretching device, amount of gel in the film in the TD stretching device, and charge on the film in the winding machine. (Postscript 8) In any of the presumption methods described in Appendix 1 to Appendix 7 The detection data selected from a variety of pre-defined detection data types will be input into the estimation model. (Note 9) In any of the presumption methods described in Appendix 1 to Appendix 8 Prepare multiple types of inference models generated by different algorithms. The estimated model to be used is selected based on the type of membrane forming machine. (Postscript 10) In any of the presumption methods described in Appendix 1 to Appendix 9 The estimation model includes multiple models, and the abnormality or lifespan of the film forming machine is estimated based on the output of each of the multiple models. Explanation of reference numerals in the attached figures

[0112] 100 Presumption System 1. Membrane forming machine 11 Extruder 12 Casting apparatus 13 MD stretching device 14 TD tensioning device 15 Winding Machine 16. Control device 2 Detectors 3. Data collection device 31 Control Department 32 Storage Department 33 Ministry of Communications 34 Data Input Section 4. Estimation device 41 Control Department 42 Storage Department 43 Ministry of Communications 4A Recording Media 4P program 421 Molding Information DB 422 Presumed Model 5a, 5b Terminal devices.

Claims

1. A method of estimation, wherein, Acquire detection data regarding the status of the membrane forming machine detected by the detector. The abnormality or lifespan of the membrane forming machine is estimated by inputting the acquired detection data into the estimation model. The estimation model estimates the abnormality or lifespan of the membrane forming machine in the presence of detection data about the state of the membrane forming machine.

2. The estimation method according to claim 1, wherein, The detection data is acquired during the continuous operation of the film forming machine. Under continuous operation, the abnormality or lifespan of the film forming machine is inferred based on the detection data.

3. The estimation method according to claim 1 or 2, wherein, Derive the feature values ​​derived from the detection data. The abnormality or lifespan of the membrane forming machine is estimated by inputting the derived feature quantities into the estimation model, which estimates the abnormality or lifespan of the membrane forming machine in the presence of feature quantities derived from detection data about the state of the membrane forming machine.

4. The estimation method according to any one of claims 1 to 3, wherein, Information indicating an estimated malfunction or lifespan of the membrane forming machine is sent to the terminal device.

5. The estimation method according to any one of claims 1 to 4, wherein, Based on the time-series detection data, information representing the trend of the detection data is generated. The generated information representing the trend of the detection data is sent to the terminal device.

6. The estimation method according to any one of claims 1 to 5, wherein, The film forming machine includes an extruder, a casting device, an MD stretching device, a TD stretching device, and a winding machine. The detection data includes the physical quantities of the extruder, the physical quantities of the casting device, the physical quantities of the MD stretching device, the physical quantities of the TD stretching device, or the physical quantities of the winding machine.

7. The estimation method according to any one of claims 1 to 6, wherein, The film forming machine includes an extruder, a casting device, an MD stretching device, a TD stretching device, and a winding machine. The detection data includes at least one of the following: acceleration in the extruder, resin viscosity in the extruder, acceleration in the casting device, gel quantity of the film in the casting device, acceleration in the MD stretching device, acceleration in the TD stretching device, wind speed in the TD stretching device, air volume in the TD stretching device, strain of the film holder in the TD stretching device, phase difference between the left and right pairs of film holders in the TD stretching device, crystal orientation in the TD stretching device, gel quantity of the film in the TD stretching device, and charge on the film in the winding machine.

8. The estimation method according to any one of claims 1 to 7, wherein, The detection data selected from a variety of pre-defined detection data types will be input into the estimation model.

9. The estimation method according to any one of claims 1 to 8, wherein, Prepare multiple types of inference models generated by different algorithms. The estimated model to be used is selected based on the type of membrane forming machine.

10. The estimation method according to any one of claims 1 to 9, wherein, The estimation model includes multiple models, and the abnormality or lifespan of the film forming machine is estimated based on the output of each of the multiple models.

11. A computer program for causing a computer to perform the following processes, wherein, The process includes: Acquire detection data regarding the status of the membrane forming machine detected by the detector. The abnormality or lifespan of the membrane forming machine is estimated by inputting the acquired detection data into the estimation model. The estimation model estimates the abnormality or lifespan of the membrane forming machine in the presence of detection data about the state of the membrane forming machine.

12. A estimation device, wherein, include: The acquisition unit acquires detection data about the state of the film forming machine detected by the detectors; and The estimation unit estimates the abnormality or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model. The estimation model estimates the abnormality or lifespan of the membrane forming machine in the presence of detection data about the state of the membrane forming machine.

13. A presumption system, wherein, It includes a membrane forming machine, a detector, and a estimation device. The estimation device includes: The acquisition unit acquires detection data regarding the state of the film forming machine detected by the detector; and The estimation unit estimates the abnormality or lifespan of the membrane forming machine by inputting the acquired detection data into an estimation model. The estimation model estimates the abnormality or lifespan of the membrane forming machine in the presence of detection data about the state of the membrane forming machine.

Images