JP2020071271A - Machine learning device, data processing system, machine learning method, and data processing method - Google Patents

Abstract

To adjust control information accurately at a low cost in printing on various print media using an image forming apparatus, and thereby obtain a desired print result.SOLUTION: A machine learning device comprises: a state variable acquisition unit 110 that acquires, as state variables, feature quantity information in an actual printed material AP on which printing is actually performed by an image forming apparatus 10, medium information on a print medium PM used for the actual printed material AP, and first control information that is control information when the actual printed material AP is output; a teacher data acquisition unit 120 that acquires, as teacher data, second control information having the feature quantity information within a predetermined threshold; and a learned model creation unit 130 that crates a learned model by performing machine learning based on the information acquired by the state variable acquisition unit 110 and the teacher data. The machine learning device executes data processing for printing by using the learned model created with a machine learning method.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a machine learning device and a machine learning method for learning a correlation between control information of an image forming device and a printed matter output by the image forming device, and a learned model obtained by the machine learning device and the machine learning method. The present invention relates to a data processing system and a data processing method using the.

  An electrophotographic image forming apparatus represented by a copying machine, a printer, and a multi-functional peripheral has been widely spread, especially in an office environment. Further, in recent years, even for the needs of printing (hereinafter, also referred to as “industry print”) specialized for a specific job in a specific industry in an industrial field such as a medical field, a manufacturing industry, or a distribution industry, An electrophotographic image forming apparatus is used.

  Examples of the industry print include packaging printing of products, label printing attached to wine bottles, printing of wedding invitations, and the like. As can be seen from the examples, in many cases, the print quality (image quality) of the printed matter in the industry print has a great influence on the value of goods and services. Therefore, print quality is particularly important for printed matter in industrial printing.

  On the other hand, industry printing is different from printing using general paper (A4 plain paper, B4 plain paper, etc.) that is usually carried out in offices and the like, and is different from a special medium such as thick paper or long paper depending on the purpose or use. Standard paper, Japanese paper, paperboard, foreign paper, film, label, envelope, etc. are used. When performing printing on such a wide variety of media (hereinafter collectively referred to as “print media”) using a common image forming apparatus, the control parameters of the image forming apparatus at the time of printing are set to the print medium. It needs to be adjusted accordingly.

  As an image forming apparatus that prints on a special print medium as described above, for example, there is one described in Patent Document 1. Japanese Patent Application Laid-Open No. 2004-163242 accepts an input of a paper selection key set for each paper according to a paper thickness of thin, medium, normal, thick, etc. from a user, and a transfer voltage and a fixing temperature corresponding to the received paper selection key. There is described an electrophotographic image forming apparatus for setting a sheet, and if the method described in Patent Document 1 is adopted, it is possible to adjust the control parameter according to the sheet thickness.

JP, 09-329994, A

  However, the paper thickness is not the only parameter that affects print quality. For example, the quality of printing varies depending on whether or not the paper is coated, the material of the paper, and the like, and the state of the image forming apparatus that actually outputs is naturally not uniform. The print quality changes depending on the condition. Therefore, it is insufficient to accurately print on various print media.

  In view of the above points, it is an object of the present invention to adjust control information with high accuracy and obtain a desired print result when printing on various print media using an image forming apparatus.

  In order to achieve the above object, the machine learning apparatus according to the first aspect of the present invention includes, as shown in FIG. 2, for example, feature amount information in an actual printed matter AP that is actually printed by the image forming apparatus 10, A state variable acquisition unit 110 that acquires, as state variables, medium information of the print medium PM used for the actual printed matter AP and first control information that is control information when the actual printed matter AP is output; A second control information that is within a threshold value of a teacher data acquisition unit 120 as teacher data; machine learning is performed based on the information acquired by the state variable acquisition unit 110 and the teacher data to generate a learned model. And a learned model generation unit 130 for

  According to this structure, the machine learning is performed by using the three pieces of feature amount information, the medium information, and the first control information as state variables, so that a desired learned model with high accuracy can be obtained.

  In the machine learning device according to the second aspect of the present invention, in the machine learning device 100 according to the first aspect of the present invention, the feature amount information includes information about a print defect in the actual printed matter AP.

  With this configuration, a learned model that can eliminate printing defects that have occurred in the actual printed matter in machine learning is generated, and a learned model that can perform data processing desired by the user can be generated.

  The machine learning device according to the third aspect of the present invention is the machine learning device 100 according to the first aspect of the present invention, wherein the medium information includes the presence / absence of coating of the print medium PM, material, thickness, and weight. And density.

  With this configuration, in machine learning, in addition to the thickness of the print medium, four types of medium information such as the presence or absence of coating are also taken into consideration, so that a learned model that sufficiently considers the medium information can be generated. it can.

  Further, a machine learning apparatus according to a fourth aspect of the present invention is, for example, as shown in FIG. 1, in the machine learning apparatus 100 according to the first aspect of the present invention, the image forming apparatus 10 is an electrophotographic type. In the image forming apparatus, the first control information includes a toner fixing temperature and a transfer voltage in the image forming apparatus 10.

  With this configuration, the learned model generated by including the two pieces of information that most affect print quality as the first control information enables the data processing desired by the user.

  Further, a machine learning apparatus according to a fifth aspect of the present invention is, for example, as shown in FIG. 1, in the machine learning apparatus 100 according to the first aspect of the present invention, the image forming apparatus 10 is an electrophotographic type. In the image forming apparatus, the second control information includes the toner fixing temperature and the transfer voltage in the image forming apparatus 10.

  With this configuration, the second control information is composed of the toner fixing temperature and the transfer voltage, which are generally used control parameters in the case of an electrophotographic image forming apparatus. The present invention can be applied to various types of electrophotographic image forming apparatuses.

  The machine learning device according to the sixth aspect of the present invention is used, for example, as shown in FIGS. 2 and 5, for the feature amount information of the actual printed matter AP actually printed by the image forming apparatus 10 and the actual printed matter AP. A state variable acquisition unit 110 that acquires, as state variables, medium information of the obtained print medium PM, first control information that is control information when the actual printed matter AP is output, and environmental information around the image forming apparatus AP; A teacher data acquisition unit 120 that acquires, as teacher data, second control information that sets the feature amount information within a predetermined threshold; learning is performed by performing machine learning based on the information acquired by the state variable acquisition unit and the teacher data. And a trained model generator 130 for generating a trained model.

  With this configuration, machine learning is performed by using four pieces of information including feature quantity information, medium information, and first control information as environment variables, and thus machine learning is performed. Can be obtained.

  In the machine learning device according to the seventh aspect of the present invention, in the machine learning device 100 according to the sixth aspect of the present invention, the environment information includes temperature and humidity around the image forming apparatus 10.

  According to this structure, the environmental information includes information on the temperature around the image forming apparatus that affects the fixing temperature control in the image forming apparatus and the humidity around the image forming apparatus that affects the state of the print medium. With this, it is possible to obtain a desired learned model with high accuracy according to the situation at the time of printing.

  The data processing system according to the eighth aspect of the present invention is, for example, as shown in FIGS. 6 and 8, the feature amount information of the actual printed matter AP actually printed by the image forming apparatuses 10 and 200, and the actual printed matter AP. An actual printed matter information acquisition unit 230 that acquires medium information of the print medium PM used for the first printed matter and first control information that is control information of the image forming apparatuses 10 and 200 when the actual printed matter AP is output; The data processing unit 260 that inputs the information acquired by the acquisition unit 230 into the learned model generated by the machine learning device 100 according to any one of the first to fifth aspects and outputs the third control information. And; a third control information storage unit 230 that stores the third control information output from the data processing unit 260.

  With this configuration, by using the learned model, it is possible to automatically adjust the control information without the need for adjustment by a technician, and thus it is possible to reduce printing defects caused on the actual printed matter at low cost. In addition, it is possible to provide data processing that can be realized with high accuracy.

  The data processing system according to the ninth aspect of the present invention is, for example, as shown in FIG. 6 and FIG. 8, feature amount information of the actual printed matter AP actually printed by the image forming apparatuses 10 and 200, and the actual printed matter AP. The medium information of the print medium PM used in the above, the first control information that is the control information of the image forming apparatuses 10 and 200 when the actual printed matter AP is output, and the environmental information around the image processing apparatuses 10 and 200 are acquired. The actual printed matter information acquisition unit 230; and the information acquired by the actual printed matter information acquisition unit 230 is input to the learned model generated by the machine learning apparatus 100 according to the sixth or seventh aspect, and the third control information is input. And a third control information storage section 253 that stores the third control information output from the data processing section 260.

  With this configuration, by using the learned model, it is possible to automatically adjust the control information without the need for adjustment by a technician, and thus it is possible to reduce printing defects caused on the actual printed matter at low cost. In addition, it is possible to provide data processing that can be realized with higher accuracy.

  The machine learning method according to the tenth aspect of the present invention uses a computer as shown in FIG. 4, for example; the feature amount information of the actual printed matter AP actually printed by the image forming apparatus 10 and the actual printed matter AP are used. Step S12 of acquiring the medium information of the used printing medium PM and the first control information which is the control information when the actual printed matter AP is output as a state variable; and the second in which the characteristic amount information is within a predetermined threshold value. Step S13 of acquiring control information as teacher data; Step S14 of generating a learned model by performing machine learning based on the feature amount information, the medium information, the first control information, and the teacher data; Is included.

  According to this structure, the machine learning is performed by using the three pieces of feature amount information, the medium information, and the first control information as state variables, so that a desired learned model with high accuracy can be obtained.

  The machine learning method according to the eleventh aspect of the present invention uses a computer, for example, as shown in FIG. 5; the feature amount information in the actual printed matter AP actually printed by the image forming apparatus 10 and the actual printed matter AP Step S22 of acquiring medium information of the used printing medium PM, first control information that is control information when the actual printed matter AP is output, and environmental information around the image forming apparatus 10 as state variables; Step S13 of acquiring, as teacher data, second control information whose information is within a predetermined threshold; machine learning is performed based on the feature amount information, the medium information, the first control information, the environment information, and the teacher data. This includes step S15 of generating a learned model by executing;

  With this configuration, machine learning is performed by using four pieces of information including feature quantity information, medium information, and first control information as environment variables, and thus machine learning is performed. Can be obtained.

  The data processing method according to the twelfth aspect of the present invention is, for example, as shown in FIG. 7, using the computer, the characteristic amount information of the actual printed matter AP actually printed by the image forming apparatuses 10 and 200, and Step S32 of acquiring the medium information of the print medium PM used for the printed matter AP and the first control information which is the control information of the image forming apparatus 10 or 200 when the actual printed matter AP is output; Step S34 of inputting the medium information and the first control information to the learned model generated by the machine learning method according to claim 10 and outputting the third control information; the output third control information Is stored in step S35.

  With this configuration, by using the learned model, it is possible to automatically adjust the control information without the need for adjustment by a technician, and thus it is possible to reduce printing defects caused on the actual printed matter at low cost. In addition, it is possible to provide data processing that can be realized with high accuracy.

  A data processing method according to a thirteenth aspect of the present invention is, for example, as shown in FIG. 7, using a computer, the characteristic amount information of the actual printed matter AP actually printed by the image forming apparatuses 10 and 200, and The medium information of the print medium PM used for the printed matter AP, the first control information that is the control information of the image forming apparatuses 10 and 200 when the actual printed matter is output, and the environmental information around the image forming apparatuses 10 and 200 are displayed. Acquiring step S32; inputting the acquired characteristic amount information, medium information, first control information, and environment information into the learned model generated by the machine learning method according to claim 11, and performing the third control. The step S34 of outputting information; the step S35 of storing the output third control information are included.

  With this configuration, by using the learned model, it is possible to automatically adjust the control information without the need for adjustment by a technician, and thus it is possible to reduce printing defects caused on the actual printed matter at low cost. In addition, it is possible to provide data processing that can be realized with higher accuracy.

  According to the present invention, the feature amount information of the actual printed matter, the medium information of the print medium used for the actual printed matter, the first control information that is the control information when the actual printed matter is output (or in addition to these, the image Since the learned model is generated by the machine learning device and the machine learning method using the environment information around the forming device) as a state variable, it is possible to generate a highly accurate learned model according to the user's request. Become. Further, by adopting the automatic adjustment of the control information using the learned model obtained here, it becomes possible to adjust the control information without requiring any human labor, and it is possible to obtain a highly accurate printing result at a low cost. It becomes possible to provide a possible data processing system and data processing method.

FIG. 1 is a schematic structural diagram of an image forming apparatus used in an embodiment of the present invention. FIG. 2 is a schematic block diagram of the machine learning device according to the embodiment of the present invention. FIG. 3 is a diagram showing an example of a neural network model for supervised learning implemented in the machine learning device according to the embodiment of the present invention. FIG. 4 is a flowchart showing a machine learning method according to an embodiment of the present invention. FIG. 5 is a flowchart showing a machine learning method according to another embodiment of the present invention. FIG. 6 is a schematic block diagram of the data processing system according to the embodiment of the present invention. FIG. 7 is a flowchart showing a data processing method executed in the data processing unit of the data processing system according to the embodiment of the present invention. FIG. 8 is a schematic block diagram of a data processing system according to another embodiment of the present invention.

  Hereinafter, an embodiment for carrying out the present invention will be described with reference to the drawings. In the following, the range necessary for the explanation to achieve the object of the present invention is schematically shown, and the range necessary for the explanation of the relevant part of the present invention will be mainly explained, and about the place where the explanation is omitted, A known technique is used.

  First, the basic configuration of an electrophotographic image forming apparatus that performs printout as a learning target of a machine learning apparatus and a machine learning method according to an embodiment of the present invention will be briefly described below.

<Image forming device>
FIG. 1 is a schematic structural diagram of an image forming apparatus used in an embodiment of the present invention. The image forming apparatus 10 illustrated here is a so-called intermediate transfer type full-color LED printer, and as shown in FIG. 1, the print medium supply unit 20, the image forming unit 30, the transfer unit 40, and the fixing unit 50. And an output unit 60 and a control unit 70.

  The print medium supply unit 20 is for supplying the print medium PM into the image forming apparatus 10, and includes a paper tray 21, a manual feed tray 22, and a plurality of paper feed rollers 23. The paper tray 21 is one in which a plurality of print media PM are stacked and accommodated. The print medium PM accommodated in the paper tray 21 is generally ordinary paper (A4 plain paper, B4 plain paper, etc.). The manual feed tray 22 is provided so as to be able to be housed on the side surface of the main body of the image forming apparatus 10, and is a tray for feeding the special print medium PM mainly when printing on a special print medium PM different from general paper. Is. Therefore, when performing the industry print, the manual feed tray 22 is mainly used. The plurality of paper feed rollers 23 are rollers arranged at appropriate positions for transporting the print medium PM in the paper tray 21 or placed on the manual feed tray 22 to the transport path PL.

  The image forming unit 30 is for forming a toner image, and includes a plurality of (five in the present embodiment) image forming units 31C, 31M, 31Y, 31K, 31S (hereinafter, these units are arranged in parallel). Collectively referred to as an image forming unit 31). The plurality of image forming units 31C, 31M, 31Y, 31K, 31S are basically the same in configuration except that the toner colors are different. Mainly, the photosensitive drum, the charging roller, and the developing roller are , An LED head and a toner tank. As for the image forming process with each of these configurations, a process similar to a well-known process can be adopted, and a detailed description thereof will be omitted here.

  The plurality of image forming units 31C, 31M, 31Y, 31K and 31S are units for forming toner images of cyan (Cyan), magenta (Yellow), black (BlacK) and special colors, respectively. By employing the plurality of image forming units 31, full color printing is possible. The toner image formed by each image forming unit 31 is transferred to an intermediate transfer belt 41 of a transfer unit 40 described later. As the special color toner, for example, white toner, clear toner, or fluorescent color (neon yellow, etc.) toner can be adopted.

  The transfer unit 40 is for transferring the toner image formed by the image forming unit 30 onto the print medium PM, and includes an intermediate transfer belt 41 and a plurality of primary transfer rollers 42C, 42M, 42Y, 42K, 42S (hereinafter , These are collectively referred to as a primary transfer roller 42), a backup roller 43, and a secondary transfer roller 44. The intermediate transfer belt 41 is an endless elastic belt which is supported by a plurality of rollers including a driving roller and is mainly made of a resin material such as rubber. On the surface of the intermediate transfer belt 41, the toner image of each color formed by the image forming unit 31 is transferred (primary transfer), and the formed toner image is later transferred to the print medium PM (secondary transfer). ) Will be done. The plurality of primary transfer rollers 42C, 42M, 42Y, 42K, 42S are for transferring the toner images of the respective colors formed by the image forming unit 31 to the intermediate transfer belt 41, and the plurality of image forming units 31C, 31M. , 31Y, 31K, 31S are arranged so as to face each other, and the intermediate transfer belt 41 is sandwiched between the facing photosensitive drums. A predetermined primary transfer voltage is applied to the primary transfer roller 42. The primary transfer voltage is controlled by the control unit 70 described later.

  The backup roller 43 is one of a plurality of rollers that supports the intermediate transfer belt 41, and is arranged at a position facing a secondary transfer roller 44 described later with the intermediate transfer belt 41 interposed therebetween. The secondary transfer roller 44 is arranged in the middle of the conveyance path PL at a position facing the backup roller 43 with the intermediate transfer belt 41 interposed therebetween. It functions to transfer the toner image previously formed on the intermediate transfer belt 41 when the print medium PM is passed during the interval. A predetermined secondary transfer voltage is applied to the secondary transfer roller 44. The secondary transfer voltage is controlled by the control unit 70 described later.

  The fixing unit 50 is for fixing the toner image on the print medium PM by applying heat and pressure to the print medium PM onto which the toner image has been transferred by the transfer unit 40, and the fixing roller 51 and the pressure roller And a roller 52. The fixing roller 51 has a heater (not shown) built therein, and the fixing temperature is controlled by the current value supplied to the heater. The current value supplied to this heater is controlled by the control unit 70 described later. The pressure roller 52 has a biasing force applied to the fixing roller 51. As a result, a predetermined fixing pressure is applied to the print medium PM passing between the fixing roller 51 and the pressure roller 52. In this embodiment, the pressure roller 52 is biased against the fixing roller 51. However, instead of the pressure roller 52, a fixed backup roller is used to fix the fixing roller 51. May be biased to the backup roller.

  The output unit 60 outputs the print medium PM having the toner image fixed by the fixing unit 50 to the outside of the image forming apparatus 10 as an actual print AP, and includes a discharge tray 61 and a plurality of conveyance rollers 62. The discharge tray 61 is formed on the upper part of the image forming apparatus 10 and on which the actual printed matter AP output via the transport path PL is placed. The plurality of transport rollers 62 are provided at various places on the transport path PL and transport the print medium PM to the discharge tray. Further, a cooling unit for removing heat generated when the toner image is fixed may be optionally provided at any position of the output unit 60. As the cooling unit, for example, at least a part of the transport roller 62 is provided. A roller having a heat radiation function may be adopted, or a well-known heat pipe, heat sink, fan or the like may be arranged as a cooling unit at a predetermined position of the output section.

  The control unit 70 is for controlling each unit of the image forming apparatus 10, and includes a well-known CPU, storage means, and the like. As described above, the control unit 70 has a function of controlling the secondary transfer voltage applied to the secondary transfer roller 44 and the current supplied to the heater inside the fixing roller 51.

  As described above, the intermediate transfer type full-color LED printer has been described as the image forming apparatus 10 used in the embodiment of the present invention, but the image forming apparatus to which the present invention is applicable is limited thereto. Not a thing. To give a concrete example of this point, a tandem type or rotary type that directly transfers from a photosensitive drum to a printing medium is used instead of the intermediate transfer type, and monochrome or special color toner is used instead of full color. Not to use full-color type, to use a laser head instead of an LED head as an exposing means, and to use a copying machine, a facsimile or a digital multi-function peripheral combining these functions instead of a printer. Etc. are possible.

  When performing the industry print using the image forming apparatus 10 having the above-described configuration, it is difficult to sufficiently adjust the control parameters in the conventional technique as described in the above problem. As a result, in industrial printing, a high rate of low-quality printing (printing failure) occurs. When such a printing defect occurs, a professional engineer comprehensively considers the state of the actual printed matter that was actually printed, the control information of the image forming apparatus when printing, etc. It is conceivable that a desired printing result is obtained by deriving optimum control parameters based on the experience and accumulated know-how. However, this series of operations requires a long time, and a technician is occupied for each occurrence of printing failure, which results in a very high cost. Therefore, the present invention attempts to automate the adjustment of control information (control parameters) of the image forming apparatus 10 by using a learned model generated by the machine learning apparatus 100 and the machine learning method described below. Is.

  By the way, when performing printing using the image forming apparatus as described above, various control parameters are adjusted in order to obtain an optimum printing result. When the present inventors have verified these various control parameters, among the various control parameters, the secondary transfer voltage applied to the secondary transfer roller 44 and the fixing temperature of the fixing roller 51 (that is, the inside of the fixing roller 51). It is known that two of the current values supplied to the heaters) are control information that directly affects the print quality, in other words, control information that has a high correlation with printing defects that occur on the printing surface. Was done. When the relationship between these two pieces of control information and print defects was examined in more detail, when the secondary transfer voltage was high, dust (print defects with white spots) was the main cause, and the secondary transfer voltage was reversed. When the transfer voltage is low, it is a cause of fading (printing defects such as light color). If the fixing temperature is high, it causes spots (printing defects with spot-like patterns). On the contrary, if the fixing temperature is low, fixing defects (printing defects due to toner peeling) and image misalignment (density) occur. It is known that this causes printing defects such as a thin portion of the image.

  Therefore, in the machine learning device 100 according to the embodiment of the present invention described below, the control information to be adjusted is the secondary transfer voltage and the fixing temperature, and learning for adjusting these two control information is performed. A concrete configuration and a series of machine learning methods for creating a completed model will be explained.

<Machine learning device>
FIG. 2 is a schematic block diagram of the machine learning device 100 according to the embodiment of the present invention. The machine learning device 100 includes a state variable acquisition unit 110, a teacher data acquisition unit 120, a learned model generation unit 130, and a storage unit 140. As can be understood from the above-described components, the machine learning device 100 according to the present embodiment is a device that generates a learned model by so-called supervised learning. 2 illustrates the machine learning device 100 built in a computer (server device, PC, or the like) separate from the image forming apparatus 10 for the purpose of facilitating the understanding. The apparatus 100 may be built in the image forming apparatus 10.

  The state variable acquisition unit 110 acquires, as a state variable, parameter information necessary for generating a learned model capable of adjusting the control information of the image forming apparatus 10. In machine learning, state variables are the most important factors that determine the accuracy of the trained model generated. Further, if the combination of the information acquired as the state variable is different, the learned model to be generated is naturally different. Therefore, it is a particularly important item in the present invention that the combination is an extremely important element.

  In the present embodiment, when the state variable acquisition unit 110 outputs the characteristic amount information of the actual printed matter AP, the medium information of the print medium PM (see FIG. 1) used for the actual printed matter AP, and the actual printed section AP. The control information (hereinafter, referred to as “first control information”) is acquired as state variables. The method of acquiring the state variable can be arbitrarily set according to the connection state between the machine learning apparatus 100 and the image forming apparatus 10, and the communication means or the like that is local or via the Internet can be used, or any storage can be performed. Acquired via the medium. Then, the acquired three pieces of information are stored in the storage unit 140 described later as one data set.

  The feature amount information in the actual printed matter AP is information about the printing defect occurring in the actual printed matter AP, and more specifically, includes information about the information about the defective printing. The print defect information is information on the degree of print defect. Here, the characteristic amount information that is actually input to the machine learning device 100 may be information of image data obtained by reading the actual printed matter AP using a well-known scanner, and may be a specific print failure in the actual printed matter AP. It is not necessary to specify the type and degree in advance. This is because the learning method of the machine learning device learns the control information as a suitable output result for the state variable regardless of the type and degree of printing failure, and it is necessary to determine even the type and degree of printing failure. This is because there is no However, it is of course possible to employ a pre-processing process for adjusting the information of the image data in advance to information suitable for input to the input layer of the machine learning device. A person skilled in the art can easily understand that a method generally used in the technical field of image recognition can be adopted as a specific pre-processing process, and therefore the description thereof is omitted here.

  The medium information of the print medium PM is various kinds of information about the print medium PM, and is preferably information regarding the presence / absence of coating of the print medium PM, material, thickness, weight, and density. The above five types of medium information, that is, the presence or absence of coating, the material, the thickness, the weight, and the density are specified by the present inventor as parameters that particularly affect the quality of printing, and based on these five types of medium information. By performing machine learning by using this method, a highly accurate learned model can be efficiently generated. The material as the medium information only needs to specify the main material used in the print medium PM, and does not necessarily require the information about the additionally included material. As the weight as the medium information, various kinds of weight can be used as long as they show characteristics related to the weight of the print medium. Specifically, for example, the basis weight or the continuous weight of the image forming apparatus can be used. What is generally used in the technical field can be utilized.

  The first control information is a control parameter that was actually set when the actual printed matter AP was output, and is preferably the toner fixing temperature of the fixing roller 51 in the image forming apparatus 10 and the secondary transfer roller 44. It is the secondary transfer voltage applied. In the image forming apparatus 10 described above, the secondary transfer voltage is used as the first control information because the toner image is transferred from the intermediate transfer belt 41 to the print medium PM. In a tandem type image forming apparatus that does not have a transfer unit and directly transfers the toner image from the photosensitive drum to the print medium, the toner image is transferred from the photosensitive drum to the print medium PM provided in each of the plurality of image forming units having different toner colors. The transfer voltage at that time may be used as the first control information.

  The teacher data acquisition unit 120 obtains, in the actual printed matter AP including the print defect, control information in which the feature amount information, which is information about the print defect, falls within a predetermined threshold, in short, a print result that does not include the print defect. Such improved control information (hereinafter referred to as second control information) is acquired as teacher data. The second control information is, for example, control information derived by an engineer based on the output result of the actual printed matter AP and the control information at that time. Therefore, the "predetermined threshold value" does not necessarily mean a specific value, and if the control information is such that an appropriate output result can be obtained from an engineer's point of view, the control information is " It can be said that the feature amount information is within a predetermined threshold ”. The second control information is acquired by the technician EN by directly inputting it to the teacher data acquisition unit or by being transmitted via various communication means or any storage medium. The second control information is stored in the storage unit 140 in association with the corresponding data set acquired by the state variable acquisition unit 110 and stored in the storage unit 140 as a data set.

  The learned model generation unit 130 is based on the data set including three pieces of information as state variables acquired by the state variable acquisition unit 110 and the teacher data acquired by the teacher data acquisition unit 120 and associated with the corresponding data set. Machine learning is performed to generate a trained model. A specific method of machine learning performed here will be described in detail below.

  The machine learning device 100 according to the present invention employs supervised learning using a neural network model as its learning method.

  FIG. 3 is a diagram showing an example of a neural network model for supervised learning implemented in the machine learning device according to the embodiment of the present invention. The neural network in the neural network model shown in FIG. 3 has l neurons (x1 to xl) in the input layer, m neurons (y11 to y1m) in the first intermediate layer, and n neurons in the second intermediate layer. Of neurons (y21 to y2n) and two neurons (z1, z2) in the output layer. The first intermediate layer and the second intermediate layer are also called hidden layers, and the neural network may have a plurality of hidden layers in addition to the first intermediate layer and the second intermediate layer. Alternatively, only the first intermediate layer may be the hidden layer.

  Further, between the input layer and the first intermediate layer, between the first intermediate layer and the second intermediate layer, and between the second intermediate layer and the output layer, nodes connecting neurons between layers are provided. The weight wi (i is a natural number) is associated with each node.

  The neural network in the neural network model according to the present embodiment uses the learning data set to learn the correlation between the control information of the image forming apparatus and the printed matter output by the image forming apparatus. Specifically, each of the three state variables is associated with a neuron in the input layer, and the value of the neuron in the output layer is calculated by a method of calculating the output value of a general neural network, that is, the value of the neuron on the output side. , The value of the input-side neuron connected to the neuron and the weight wi associated with the node connecting the output-side neuron and the input-side neuron are calculated as the sum of the sequence of multiplication values, It is calculated by using the method performed for all neurons other than the neurons in the input layer. When associating the state variables with the neurons in the input layer, the format in which the information acquired as the state variables is associated can be appropriately set in consideration of the accuracy of the learned model generated. For example, when associating the image data of the actual printing portion AP as the characteristic amount information in the actual printed matter AP with the input layer, the image data is divided into predetermined areas, and then the color value (for example, RGB value) of each divided area is divided. ) Information can be associated with each input layer.

  Then, the calculated values of the two neurons z1 and z2 in the output layer, that is, the toner fixing temperature of the fixing roller 51 and the secondary transfer voltage applied to the secondary transfer roller 44 in this embodiment, and the data set. And the teacher data t1 and t2, which are also associated with the toner fixing temperature of the fixing roller 51 and the secondary transfer voltage applied to the secondary transfer roller 44, are compared with each other to obtain an error. The adjustment of the weight wi associated with each node so as to be small (back pro vacation) is repeated.

  Then, if a predetermined condition such as repeating the above-described series of steps a predetermined number of times or if the error becomes smaller than an allowable value is satisfied, the learning is terminated and the neural network model (of All weights wi) associated with each of the nodes are stored in the storage unit 140 as learned models.

  The learned model stored in the storage unit 140 is applied to an actual system via a communication means such as the Internet or a storage medium in response to a request. A specific application mode of the learned model to the real system (data processing system) will be described later in detail.

<Machine learning method>
In connection with the machine learning device described above, the present invention also provides a machine learning method. FIG. 4 is a flowchart showing a machine learning method according to an embodiment of the present invention. This machine learning method is realized by using a computer, but various kinds of computers can be applied, for example, the one forming the control unit 70 in the image forming apparatus 10 or the image forming apparatus 10. There may be mentioned a PC locally connected to the server or a server device arranged on the network.

  When performing supervised learning as the machine learning method according to the present invention, first, a pre-learning model having weights of initial values is prepared (step S11). Next, the characteristic amount information of the actual printed matter AP actually printed by the image forming apparatus 10, the medium information of the print medium PM used for the actual printed matter AP, and the first control information are acquired as the state variables (step). S12). Then, while acquiring the teacher data corresponding to the state variable acquired in step S12 (step S13), or before or after this, the state variable acquired in step S12 is input to the pre-learning model (see FIG. 3). The output layer (see FIG. 3) is generated by associating with (step S14).

  The control information that forms the output layer generated in step S14 is generated by the pre-learning model, and thus is not control information that normally provides a print result that satisfies the user's request. Therefore, next, machine learning is carried out using the control information forming the teacher data acquired in step S13 and the control information forming the output layer generated in step S14 (step 15). The machine learning performed here is to compare the control information forming the teacher data with the control information forming the output layer, detect an error between them, and perform learning so as to obtain an output layer that reduces this error. Machine learning is performed by adjusting the weights associated with each node in the previous model.

  When the machine learning is performed in step S15, it is determined whether or not the machine learning needs to be further performed (step S16), and when the machine learning is continued (No in step S16), the process returns to step 12, When the machine learning is to be ended (Yes in step S16), the process proceeds to step S17. When the machine learning is continued, the steps S12 to S15 are performed a plurality of times, and normally, the accuracy of the learned model finally generated increases in proportion to the number of times.

  When ending the machine learning, the neural network generated by adjusting the weights associated with each node through a series of steps is stored in the storage unit 140 as a learned model (step S17), and the series of learning processes is performed. finish. The learned model stored here is applied to and used in various data processing systems, the details of which will be described later.

  In the learning process and the machine learning method of the machine learning device described above, in order to generate one learned model, the machine learning process is repeatedly executed for one neural network (pre-learning model). Although the method for improving the accuracy and generating the trained model to be applied to the data processing system is described, the present invention is not limited to this acquisition method. For example, a plurality of learned models that have been subjected to machine learning a predetermined number of times are stored in the storage unit 140 as one candidate, and a data set for validity determination is input to the plurality of learned models to output the output layer. It is also possible to generate (value of neuron of), compare and examine the accuracy of the control information specified in the output layer, and select one best learned model to be applied to the data processing system. The validity determination data set may have the same state variables as the learning data set used for learning, and may be composed of different data.

  As an experimental example, a trained model obtained after repeatedly performing the series of learning processes shown in steps S12 to S16 for 15000 times while changing the state variables to be acquired is prepared, and 1500 patterns are input to the input layer of this trained model. The control information of the output layer obtained by inputting the state variables of the above is compared with the control information specified by the engineer based on each of the 1500 state variables, and the validity of the degree of the error is judged. Carried out. As a result, the ratio in which the control information of the output layer output by the learned model and the control information specified by the engineer were substantially the same was about 90%. This proves that the three state variables selected in this embodiment are the most important information that affects the print result.

  As described above, according to the machine learning device and the machine learning method according to the embodiment of the present invention, particularly, printing defects that occur when printing on various printing media such as industry printing can be performed with high accuracy. An improved trained model can be generated. Therefore, by applying this learned model to an actual system, it becomes possible to improve printing defects with high accuracy and obtain a desired printing result. Also, since the print information can be automatically adjusted by using the learned model, it is possible to realize a low-cost data processing system without requiring a manpower each time a print defect occurs. In addition, due to this cost reduction, users can easily try out a plurality of special print media during industry printing without being aware of the cost, so that the user has more freedom in selecting print media. Is improved. Further, by specifying the state variables input to the input layer of the machine learning device and the machine learning method to the above-mentioned three pieces of information, it is possible to perform highly accurate learning completion applicable to printout on various special print media. It is possible to generate a model with high efficiency.

<Other Embodiments>
In the above-described embodiment, three pieces of information, that is, the characteristic amount information of the actual printed matter AP, the medium information of the print medium PM used for the actual printed matter AP, and the first control information when the actual printing section AP is output, are provided. Although the machine learning device and the machine learning method that are acquired as state variables have been described, as a result of further study by the present inventor, in addition to the above three state variables, environmental information around the image forming apparatus 10 is added as a state variable. It was confirmed that a highly accurate trained model can be generated also by acquiring with. Therefore, as another embodiment, a machine learning method when the above four pieces of information are selected as state variables will be described below. Note that the configuration of the machine learning device according to another embodiment shown below is the same as that described in the above one embodiment except that the information acquired by the state variable acquisition unit 110 is different. A detailed description is omitted.

  FIG. 5 is a flowchart showing a machine learning method according to another embodiment of the present invention. As can be seen from FIG. 5, the series of steps in the present embodiment are almost the same as the steps of the machine learning method according to the above-mentioned one embodiment except step S22. Therefore, the same steps will be denoted by the same reference numerals and the description thereof will be cited here, and the portions different from the one embodiment will be mainly described.

  In the machine learning method according to the present embodiment, after the pre-learning model is prepared, the feature amount information of the actual printed matter AP actually printed by the image forming apparatus 10 and the print medium PM used for the actual printed matter AP are obtained. The medium information, the first control information, and the environmental information around the image forming apparatus 10 are acquired as state variables (step S22).

  The environmental information is information including temperature and humidity around the position where the image forming apparatus 10 is installed. As a specific method for measuring temperature and humidity, a sensor for measuring temperature and humidity is installed in the image forming apparatus 10, or output of a temperature and humidity sensor provided separately from the image forming apparatus 10. It can be measured by obtaining Further, instead of this measurement operation, the temperature and humidity around the image forming apparatus 10 may be acquired by a user's input operation directly or via a communication means.

  When the state variable is acquired in step S22, the teacher data corresponding to this state variable, that is, the teacher data composed of four pieces of information is acquired (step S13), or before or after this, the state acquired in step S22. An output layer is generated using the variables (step S14). Then, machine learning is performed based on the output layer and the teacher data obtained in the above steps (step S15), and a series of learning processes are executed a plurality of times to generate a learned model.

  According to the machine learning method described above, the number of state variables to be acquired is increased by one as compared with the above-described embodiment, so that the environment in which the image forming apparatus 10 is placed (for example, in a cold region). Machine learning that considers even whether it is placed in a dry room, etc.) can be realized, and the trained model generated here can adjust the control information with higher accuracy. Realization can be expected.

  As described above, the machine learning apparatus 100 may be incorporated in the image forming apparatus 10 or a normal PC as an application form. However, in machine learning, particularly when a large number of input parameters exist, Since the amount of calculation is extremely large, it may take a long time to generate the learned model with only the CPU mounted in the normal image forming apparatus 10. Therefore, when the machine learning device 100 is built in the image forming device 10 or a normal PC, a high-performance arithmetic device such as a GPU is added, or another PC connected to perform a machine learning operation via a network. It is preferable to take measures for shortening the time of the arithmetic processing, such as utilizing the arithmetic capacity of the server device.

  As described above, the selection of the state variable in the machine learning device of the present invention is an important factor that directly affects the learned model to be generated, but the three variables adopted in the above-described embodiment are used. , Or any combination other than the four combinations is not intended to be completely eliminated. For example, a machine learning device or the like to which an input parameter that has a sufficiently small effect on the generated learned model is added compared to the three or four parameters adopted in the above-described embodiment is substantially the same. Since it does not deviate from the technical idea of the invention, it can be said to be included in the technical scope of the present invention.

<Data processing system and data processing method>
Next, an application example of the learned model generated by the above machine learning device and machine learning method will be explained. FIG. 6 is a schematic block diagram of the data processing system according to the embodiment of the present invention. In the present embodiment, as a data processing system, a case where a trained model generated through the machine learning method according to the other embodiment is applied to the image forming apparatus 200 will be described as an example. ..

  The image forming apparatus 200 according to the present embodiment is a so-called intermediate transfer type full-color LED printer similar to that described in FIG. 1, and its mechanical structure is substantially the same as that shown in FIG. Therefore, in the following description, as for the mechanical structure of the image forming apparatus 200, the reference numerals used in the image forming apparatus 10 are referred to as the same as those of the image forming apparatus 10 as necessary. As shown in FIG. 6, the image forming apparatus 200 includes a display / operation unit 210, an output control unit 220, an actual printed matter information acquisition unit 230, a temperature / humidity sensor 240, a storage unit 250, and data. And a processing unit 260.

  The display / operation unit 210 includes a display unit including a liquid crystal panel or the like provided at a predetermined position of the image forming apparatus 200 and an operation unit including an operation button or a touch panel, and is predetermined for a user who uses the image forming apparatus 200. This is a configuration for performing notification of and enabling an input operation. The output control unit 220 is a configuration for controlling various configurations for realizing print output by the image forming apparatus 200, and is exemplified by one including at least a CPU and a memory such as a RAM and a ROM. .. The control by the output control unit 220 includes controlling the secondary transfer voltage applied to the secondary transfer roller 44 and the fixing temperature of the fixing roller 51 (specifically, the current supplied to the heater inside the fixing roller 51). I'm out.

  The actual printed matter information acquisition unit 230 is a configuration for acquiring information related to the actual printed matter AP including printing defects, and is a feature amount information acquisition unit 231, a medium information acquisition unit 232, and a first control information acquisition unit 233. And an environment information acquisition unit 234. The various types of information acquired by the actual printed matter information acquisition unit 230 are associated with each other as one data set and stored in the data set storage unit 252 described later.

  The characteristic amount information acquisition unit 231 has a configuration for acquiring image data of the actual printed matter AP as the characteristic amount information. As shown in FIG. 6, the feature amount information acquisition unit 231 is locally connected to the scanner SC outside the image forming apparatus 200 via an input / output interface (not shown), and is read by the scanner SC. The image data of the printed matter AP is acquired as the characteristic amount information. In addition, the medium information acquisition unit 232 is a configuration for acquiring information about the print medium PM, more specifically, information regarding the presence or absence of coating of the print medium PM, material, thickness, weight, and density. These pieces of medium information can be acquired, for example, by the user inputting a product code previously given to the print medium PM via the display / operation unit 210. The density as the medium information is information that can be specified by calculation if the values of the thickness and the weight (basis weight) can be specified, and thus such an input operation is not always necessary.

  The first control information acquisition unit 233 also controls the first control information, which is the control information when the actual printed matter AP is output, specifically, the secondary transfer voltage applied to the secondary transfer roller 44 and the fixing temperature of the fixing roller 51. This is a configuration for obtaining and. The first control information is information stored in the memory of the output control unit 220 or the storage unit 250 in which the control information of the output control unit 220 is stored, and thus may be acquired via the internal bus. Further, the environment information acquisition unit 234 is a configuration for acquiring environment information around the image forming apparatus 200, specifically, temperature and humidity. The environmental information can be acquired by using, for example, the temperature / humidity sensor 240 built in the image forming apparatus 200.

  The storage unit 250 constitutes a storage area for storing various information in the image forming apparatus 200, and has a learned model storage unit 251, a data set storage unit 252, and a third control information storage unit 253. Is included. The learned model storage unit 251 stores the learned model generated by the machine learning method according to the other embodiment described above, that is, four pieces of information of feature amount information, medium information, first control information, and environment information. Machine learning is performed as a variable, and the generated learned model is stored. The data set storage unit 252 stores the feature amount information, the medium information, the first control information, and the environment information regarding the common real printed matter AP acquired by the real printed matter information acquisition unit 230 as one data set. is there. The third control information acquisition unit 253 stores the third control information output by the data processing unit 260 described below.

  The data processing unit 260 uses the data set regarding the specific actual print AP stored in the data set storage unit 252 and the learned model stored in the learned model storage unit 251, and uses the input layer ( By inputting various information constituting the data set to each neuron of the second transfer voltage, the secondary transfer voltage applied to the secondary transfer roller 44 as the predetermined control information (third control information) and the fixing temperature of the fixing roller 51. And are output to the output layer. The secondary transfer voltage and the fixing temperature, which form the third control information, are control information capable of improving the print failure occurring in the actual printed matter AP. That is, data processing is performed so that a desired printed matter can be obtained by using the print medium PM used in the actual printed matter AP and performing print output with the image forming apparatus 200 using the third control information as a control parameter. The control information is adjusted by the unit 260.

  Regarding the above-described data processing system, a data processing method executed by the data processing unit 260 will be described below with reference to FIG. 7. FIG. 7 is a flowchart showing a data processing method executed in the data processing unit of the data processing system according to the embodiment of the present invention.

  As shown in FIG. 7, the data processing unit 260 maintains the standby state when no print failure has occurred (No in step S31), and the following processing is performed when the print failure has occurred (Yes in step S31). Data processing is executed. Whether or not a print defect has occurred is determined based on a user's report, and specifically, a report indicating that the print defect should be improved is issued, for example, via the display / operation unit 210 of the image forming apparatus 200, or an image. It is determined by being performed for customer service or the like that manages the forming apparatus 200. In the present embodiment, since various configurations are built in so that the printing failure can be resolved by the image forming apparatus 200 alone, if the image forming apparatus 200 is equipped with a specific mode such as a printing failure elimination mode. It is preferable in terms of convenience. In this case, when the user selects the mode via the display / operation unit 210, the following processing for eliminating the print defect is executed.

  When a print defect occurs, the image forming apparatus 200 requests the user for actual printed matter information necessary to eliminate the print defect via a predetermined user interface such as the display / operation unit 210. The actual printed matter information input by the user based on this request is stored in the data set storage unit 252 as one data set, and the data processing unit 260 refers to the data set storage unit 252 to refer to the actual printed matter information. Is acquired (step S32). As a request for the actual printed matter information to the user, various information acquisition methods are displayed on the display / operation unit 210, or navigation information is output via a voice output means (not shown) to input various information to the user. It encourages work.

  When the user inputs the necessary actual printed matter information, the data processing unit 260 refers to the learned model storage unit 251 and acquires the learned model (step S33). Then, the actual print information acquired in step S32 is input to the input layer of the learned model acquired here to generate the third control information as an output layer (step S34).

  When the third control information is generated, the data processing unit 260 temporarily stores the third control information in the third control information storage unit 253 (step S35) and waits for a reprint instruction from the user (step S36). Then, for example, when a reprint instruction is given by the user operating the display / operation unit 210, the output control unit 220 replaces the first control information with the third control information in the third control information storage unit 253. After the secondary transfer voltage applied to the secondary transfer roller 44 and the fixing temperature of the fixing roller 51 are adjusted, the printing process is executed again (step S37).

  The printed matter that has been printed and output again through the above steps has been output according to the third control information that has been adjusted based on the previous printing result, and therefore the printing failure that occurred in the previously output actual printed matter AP. Has been mostly resolved. As described above, the data processing system of the present invention does not require any manpower such as an engineer between the time when a printing failure occurs and the time when the user obtains the printed matter in which the printing failure is resolved. Therefore, in the data processing system of the present invention, it is possible to realize data processing at a low cost. Incidentally, in the printed matter reprinted through the above series of steps, in most cases, it is possible to obtain a printing result in which the printing defect is eliminated, but there is a slight possibility that the printing defect will occur even after passing through this process. is there. In such a case, the steps shown in steps S32 to S37 may be executed again.

  In addition, the information acquisition method in each configuration of the actual printed matter information acquisition unit 230 described above is not limited to the above method. As a specific example, the feature amount information acquisition unit 231 uses an imaging unit other than a scanner (for example, a camera function built in a smartphone or an image reading sensor installed in a discharge port or the like of an image forming apparatus). The characteristic amount information may be acquired by receiving the read image data of the actual printed matter AP via the Internet or the like. Further, the medium information acquisition unit 232 inputs the information or the like acquired by the user through an interactive process through the display / operation unit 210 or a predetermined application instead of the product code of the print medium PM, or By adopting various sensors for acquiring the medium information of the print medium PM on the image forming apparatus 200 side, the medium information may be acquired by a method such as automatically acquiring inside the image forming apparatus 200. .. Further, the environment information acquisition unit 234 acquires the output of a temperature / humidity sensor separately provided outside the image forming apparatus 200 instead of the temperature / humidity sensor 240 built in the image forming apparatus 200, or by the user. Environmental information may be acquired by acquiring temperature / humidity information by an input operation.

  Further, in the above-described embodiment, the image forming apparatus 200 is described as including the environment information acquisition unit 234 and the temperature / humidity sensor 240, but it is not necessary to have these configurations. However, in that case, the information acquired by the actual printed matter information acquisition unit 230 is only three pieces of information of the characteristic amount information, the medium information, and the first control information. The learned model stored in the model storage unit 251 is a learned model generated by the machine learning method according to the one embodiment, that is, the feature amount information, the medium information, and the first control information. A person skilled in the art can clearly understand that the learned model that has been machine-learned by using one piece of information as a state variable needs to be stored and acquired by the data processing unit 260.

  Further, in the above embodiment, the image forming apparatus forming the data processing system is an intermediate transfer type full-color LED printer, but instead of the printer, functions other than the printer, such as a scanner function and It may be a digital multi-function peripheral further having a facsimile function. In this case, since the digital multi-function peripheral itself has a scanner function, it is not necessary to use the external scanner SC described above when acquiring the characteristic amount information. Therefore, in this case, the data processing system according to the present embodiment can be configured even in a completely offline digital multi-function peripheral.

<Another embodiment>
The image forming apparatus 200 as the data processing system according to the above-described embodiment includes the learned model storage unit 251 inside, and the data processing for eliminating the printing failure is executed almost inside the image forming apparatus 200. However, the present invention is not limited to such a form. Therefore, an embodiment in which a series of data processing is performed outside the image forming apparatus will be described below. In the embodiments described below, only parts different from the data processing system according to the above-described embodiment will be described, and description of common configurations and functions will be omitted.

  FIG. 8 is a schematic block diagram of a data processing system according to another embodiment of the present invention. As shown in FIG. 8, the data processing system according to the present embodiment includes a server device 300 connected to the Internet. Further, a plurality of terminal devices TD (for example, smartphones, tablet terminals, PCs, etc.) and the image forming apparatus 10 are connected to the server device 300 via the Internet.

  The server device 300 includes a transmission / reception unit 310, a storage unit 320, a data processing unit 330, and a learned model correction unit 340. The transmission / reception unit 310 acquires actual printed matter information and feedback information transmitted from the terminal device TD and / or the image forming device 10, and transmits third control information described later to the terminal device TD and / or the image forming device 10. It is for. Various methods can be adopted as a method of transmitting the actual printed matter information to the transmission / reception unit 310. For example, the feature amount information and the medium information are transmitted via an application preinstalled in the terminal device TD, and the first control information (and (Environmental information) may be transmitted by the image forming apparatus 10 that operates using an operation of an application in the terminal device TD as a trigger. The third control information transmitted by the transmitting / receiving unit 310 may be transmitted to the image forming apparatus 10 so that the image forming apparatus 10 reflects the third control information, or is transmitted to the terminal device TD. The user may input the information received by the terminal device TD into the image forming apparatus 10 to reflect the third control information in the image forming apparatus 10.

  The storage unit 320 constitutes a storage area for storing various kinds of information in the server device 300, and includes a learned model storage unit 321, a data set storage unit 322, and a third control information storage unit 323. Contains. The learned model storage unit 321 stores a plurality of learned models so as to be compatible with various actual print information and various image forming apparatuses. Here, it is preferable to prepare a plurality of learned models corresponding to the types of image forming apparatuses so as to correspond to a plurality of types of image processing apparatuses having different printing methods and functions. The data set storage unit 322 stores the feature amount information, medium information, and first control information (and environment information) related to the common actual printed matter AP received by the transmission / reception unit 310 as one data set. The third control information acquisition unit 323 stores the third control information output from the data processing unit 330, which will be described later.

  The data processing unit 330 stores the data set of the specific data set transmitted via the transmission / reception unit 310 and stored in the data set storage unit 322 and the trained model among the plurality of trained models stored in the trained model storage unit 251. By using one learned model specified based on the content and the type of the image forming apparatus and the like, and by inputting various information configuring the data set to the input layer of this learned model, predetermined control information (first (3 control information), the transfer voltage and the fixing temperature are output.

  The learned model correction unit 340 is for correcting the corresponding learned model in the learned model storage unit 321 based on the feedback information received by the transmission / reception unit 310 so as to have higher accuracy. The feedback information is the terminal device TD when the printing failure is not improved or another printing failure occurs as a result of printing again based on the third control information output by the data processing unit 330. And / or information transmitted from the image forming apparatus 10. The learned model correction unit 340 uses the feedback information as a data set for learning for the corresponding learned model to correct the learned model once stored in the learned model storage unit 321 at any time. It is possible.

  The data processing method by the server device 300 according to the present embodiment is substantially the same as the method shown in FIG. 7 except that communication via the Internet is used when acquiring the actual print information and applying the third control information. Therefore, the description is omitted here.

  As described above, in the data processing system according to the present embodiment, since the data processing is realized by the server device 300, the present system can be easily applied to the existing image forming apparatus. .. Further, since the server device 300 includes the learned model correction unit 340, the learned model can be updated at any time, and the accuracy of data processing is constantly improved, so that the optimum data processing result is always provided. You can

  Note that the data processing system according to the present embodiment is described using a single server device 300 for convenience of description, but the number of server devices is not limited. It is also possible to provide this data processing system in the form of a cloud service.

  The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention. And, all of them are included in the technical idea of the present invention.

10, 200 Image forming device 44 Secondary transfer roller 51 Fixing roller 100 Machine learning device 110 State variable acquisition unit 120 Teacher data acquisition unit 130 Learned model generation unit 140, 250, 320 Storage unit 210 Display / operation unit 220 Output control unit 230 actual printed matter information acquisition unit 231 feature amount information acquisition unit 232 medium information acquisition unit 233 first control information acquisition unit 234 environment information acquisition unit 251, 321 learned model storage unit 252, 322 data set storage unit 253, 323 third control Information storage unit 260, 330 Data processing unit 300 Server device 310 Transmitter / receiver unit 340 Learned model correction unit PM print medium AP actual printed matter EN engineer SC scanner TD terminal device

Claims (13)

A machine learning device:
The feature amount information of the actual printed matter actually printed by the image forming apparatus, the medium information of the print medium used for the actual printed matter, and the first control information that is the control information when the actual printed matter is output are provided. A state variable acquisition unit that acquires as a state variable;
A teacher data acquisition unit that acquires, as teacher data, second control information that sets the feature amount information within a predetermined threshold value;
A learned model generation unit that generates a learned model by performing machine learning based on the information acquired by the state variable acquisition unit and the teacher data.
Machine learning device. The feature amount information includes information on a print defect in the actual printed matter,
The machine learning device according to claim 1. The medium information includes presence or absence of coating of the print medium, material, thickness, weight and density,
The machine learning device according to claim 1. The image forming apparatus is an electrophotographic image forming apparatus,
The first control information includes a toner fixing temperature and a transfer voltage in the image forming apparatus,
The machine learning device according to claim 1. The image forming apparatus is an electrophotographic image forming apparatus,
The second control information includes a toner fixing temperature and a transfer voltage in the image forming apparatus,
The machine learning device according to claim 1. A machine learning device:
Characteristic amount information in the actual printed matter actually printed by the image forming apparatus, medium information of the print medium used in the actual printed matter, first control information that is control information when the actual printed matter is output, A state variable acquisition unit that acquires environment information around the image forming apparatus as a state variable;
A teacher data acquisition unit that acquires, as teacher data, second control information that sets the feature amount information within a predetermined threshold value;
A learned model generation unit that generates a learned model by performing machine learning based on the information acquired by the state variable acquisition unit and the teacher data.
Machine learning device. The environmental information includes temperature and humidity around the image forming apparatus,
The machine learning device according to claim 6. Characteristic amount information of the actual printed matter actually printed by the image forming apparatus, medium information of a print medium used for the actual printed matter, and control information of the image forming apparatus when the actual printed matter is output. 1 actual printed matter information acquisition unit for acquiring control information;
Data for inputting the information acquired by the actual printed matter information acquisition unit into the learned model generated by the machine learning device according to claim 1, and outputting the third control information. With the processing unit;
A third control information storage unit that stores the third control information output from the data processing unit;
Data processing system. Characteristic amount information of the actual printed matter actually printed by the image forming apparatus, medium information of a print medium used for the actual printed matter, and control information of the image forming apparatus when the actual printed matter is output. 1 control information and an actual printed matter information acquisition unit that acquires environment information around the image processing apparatus;
A data processing unit that inputs the information acquired by the actual printed matter information acquisition unit into the learned model generated by the machine learning device according to claim 6 or 7, and outputs the third control information;
A third control information storage unit that stores the third control information output from the data processing unit;
Data processing system. A computer-based machine learning method:
The feature amount information of the actual printed matter actually printed by the image forming apparatus, the medium information of the print medium used for the actual printed matter, and the first control information that is the control information when the actual printed matter is output are provided. Obtaining as a state variable;
Acquiring second control information that sets the feature amount information within a predetermined threshold value as teacher data;
Generating a learned model by performing machine learning based on the characteristic amount information, the medium information, the first control information, and the teacher data.
Machine learning methods. A computer-based machine learning method:
Characteristic amount information in the actual printed matter actually printed by the image forming apparatus, medium information of the print medium used in the actual printed matter, first control information that is control information when the actual printed matter is output, Acquiring the environment information around the image forming apparatus as a state variable;
Acquiring second control information that sets the feature amount information within a predetermined threshold value as teacher data;
Generating a learned model by performing machine learning based on the feature amount information, the medium information, the first control information, the environment information, and the teacher data.
Machine learning methods. Using the computer, the characteristic amount information of the actual printed matter actually printed by the image forming apparatus, the medium information of the print medium used for the actual printed matter, and the image forming apparatus of the image forming apparatus when the actual printed matter is output. Obtaining first control information, which is control information;
Inputting the acquired characteristic amount information, the medium information, and the first control information to a learned model generated by the machine learning method according to claim 10, and outputting third control information;
Storing the output third control information.
Data processing method. Using the computer, the characteristic amount information of the actual printed matter actually printed by the image forming apparatus, the medium information of the print medium used for the actual printed matter, and the image forming apparatus of the image forming apparatus when the actual printed matter is output. Obtaining first control information, which is control information, and environmental information around the image forming apparatus;
The acquired characteristic amount information, the medium information, the first control information, and the environment information are input to a learned model generated by the machine learning method according to claim 11, and third control information is input. Output step;
Storing the output third control information.
Data processing method.

Images