Noise control device, image forming apparatus, and noise control system

Abstract

The present invention provides a noise control device, an image forming apparatus, and a noise control system that can enhance the noise reduction effect of an image forming apparatus. [Solution] The image forming apparatus 100 includes an estimation processing unit 405, which is a noise prediction means that predicts the noise during operation of the image forming apparatus based on machine operating conditions, which are the operating conditions of the image forming apparatus; an inverse phase waveform generation means that generates an inverse phase waveform of the predicted noise predicted by the estimation processing unit 405; and a speaker 1206, which is a sound wave output means that outputs sound waves of the inverse phase waveform generated by the operation sound reduction processing unit 411. The estimation processing unit 405 predicts the noise generated by a trained model, which is a noise prediction program that has been machine-learned using a plurality of training data.

Description

Technical Field

[0001] The present invention relates to a noise control device, an image forming apparatus, and a noise control system.

Background Art

[0002] Conventionally, there is known a noise control device having noise prediction means for predicting generated noise based on the operating conditions of an image forming apparatus, and inverse-phase waveform generation means for generating an inverse-phase waveform having an inverse phase to the noise predicted by the noise prediction means.

[0003] Patent Document 1 describes a device that predicts the noise of a driving device of an image forming apparatus as the above-described noise control device. Specifically, a data table in which each drive control command for controlling the driving device is associated with the predicted noise of the driving device corresponding to each drive control command is stored in the drive noise characteristic storage means, and based on the drive control command, the predicted noise of the driving device is specified from the drive noise characteristic storage means, thereby predicting the noise of the generated driving device. Then, based on the predicted noise of the driving device, a sound wave having an inverse phase to the predicted noise is output, and it is described that the noise of the driving device is reduced (suppressed).

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, during the operation of the image forming apparatus, the noise is a large noise due to the overlapping of various sounds such as paper conveyance sound and fan rotation sound in addition to the driving device noise. Therefore, as described in Patent Document 1, there is a problem that the noise cancellation effect of the image forming apparatus is low only by reducing the noise of the driving device.

Means for Solving the Problems

[0005] To solve the above-mentioned problems, the present invention provides a noise control device comprising: noise prediction means for predicting noise during operation of an image forming apparatus based on the operating conditions of the image forming apparatus; inverse phase waveform generation means for generating an inverse phase waveform that is in the opposite phase to the predicted noise predicted by the noise prediction means; and sound wave output means for outputting sound waves of the inverse phase waveform generated by the inverse phase waveform generation means, wherein the noise prediction means predicts the noise generated by a noise prediction program that has been machine-learned using a plurality of training data. [Effects of the Invention]

[0006] According to the present invention, the noise reduction effect of an image forming apparatus can be enhanced. [Brief explanation of the drawing]

[0007] [Figure 1] A functional block illustrating an example of a noise control system applied to an image forming apparatus. [Figure 2] Schematic diagram of an image forming apparatus. [Figure 3] A diagram showing an example of the hardware configuration of an image forming apparatus. [Figure 4] A diagram showing an example of a machine learning server hardware configuration. [Figure 5] A diagram showing an example of the software configuration of a noise control system. [Figure 6] Schematic diagram of an image forming apparatus equipped with a data collection and provision unit, a machine learning unit, and a training data generation unit. [Figure 7] A diagram showing an example of the software configuration of the image forming apparatus. [Figure 8] A conceptual diagram illustrating an example of the input / output structure using a training model in the machine learning department. [Figure 9] A diagram illustrating machine learning in the Machine Learning Department. [Figure 10] A diagram illustrating noise control during the operation of an image forming apparatus. [Figure 11] A flowchart for noise control. [Figure 12]A schematic diagram of a noise control system in which the estimation processing unit is located outside the image forming apparatus. [Modes for carrying out the invention]

[0008] The best mode for carrying out the present invention will be described below with reference to the drawings. It should be noted that a person skilled in the art will readily be able to modify or alter the present invention within the scope of the claims to create other embodiments, and such modifications and alterations are included within the scope of these claims. The following description is an example of the best mode in this invention and does not limit the scope of these claims.

[0009] The following describes embodiments of the present invention applied to an image forming apparatus. Figure 1 shows a functional block illustrating an example of a noise control system applied to an image forming apparatus. The noise control system 1 includes an image forming device 100 such as a printer, multifunction device, or fax machine, external devices 300 (machine learning server 102, data server 105), and a general-purpose computer 103 that transmits print data to the image forming device 100. These devices are connected by a network such as a LAN. The network connecting them may be wired or wireless, and the machine learning server 102 and data server 105 may be configured to communicate with the image forming device 100 via the Internet or the like.

[0010] Figure 2 is a schematic diagram of the image forming apparatus 100. The image forming apparatus 100 includes a paper feeding unit 101 which contains recording paper as a recording medium on which an image is formed, and an image forming unit 106 which serves as an image forming means. If the image forming apparatus 100 is a copier, for example, an image reading device is located on top of the main body of the image forming apparatus.

[0011] The image-forming unit 106 includes, for example, an exposure unit as an exposure means, a plurality of photosensitive drums, a developing device that uses four colors of toner: cyan (C), magenta (M), yellow (Y), and black (K), a transfer belt as an intermediate transfer body, and a secondary transfer unit. Based on the print data transmitted from a general-purpose computer 103 (see Figure 1), the image-forming unit 106 creates an image, for example, as follows: Based on the print data, the image-forming unit 106 exposes each color of photosensitive drum with the exposure unit to form an electrostatic latent image on each photosensitive drum, and develops the image by supplying toner onto the latent image on each photosensitive drum with the developing unit for each color of the developing device. The image-forming unit 106 also performs primary transfer of the toner images on each color of photosensitive drum to the transfer belt, and secondary transfer by superimposing them onto recording paper as a recording medium in the secondary transfer unit. The toner images that have been secondary transferred onto the recording paper are heated and pressurized by the fixing unit 107 to fix them onto the recording paper and form a color image. The recording paper on which the image has been formed is ejected into the output tray. Instead of the electrophotographic image forming unit described above, an image forming unit employing another recording method, such as an inkjet method, may be used.

[0012] Furthermore, the image forming apparatus 100 of this embodiment is equipped with an AI (Artificial Intelligence) function, which is used to predict the noise generated during the operation of the image forming apparatus 100 (during image forming operation), and to generate a waveform with the opposite phase to the predicted noise, thereby providing noise control means to reduce the noise during the operation of the image forming apparatus.

[0013] The machine learning server 102, shown in Figure 1, plays a central role in generating the trained model, which is a noise prediction program for realizing this AI function. The data server 105 is responsible for collecting training data used for machine learning in the machine learning server 102 from external devices such as the image forming apparatus 100 and providing it to the machine learning server 102.

[0014] The image forming apparatus 100 can receive a learned model generated by the machine learning server 102 at any time and can realize a specific AI function using the learned model. Further, the machine learning server 102 shown in FIG. 1 receives learning data necessary for learning a learning model for realizing a specific AI function from a data server 105, the image forming apparatus 100, a general-purpose computer 103, etc. Then, the machine learning server 102 generates a learned model by performing learning processing using some or all of them.

[0015] FIG. 3 is a diagram showing an example of the hardware configuration of the image forming apparatus 100. The image forming apparatus 100 includes a CPU (Central Processing Unit) 1201, a RAM (Random Access Memory) 1202, and a ROM (Read Only Memory) 1203. Further, it has a hardware group including an HDD (Hard Disk Drive) 1204 as storage means, a GPU (Graphics Processing Unit) 1221, a network IF 1210 as a transmission / reception unit, and various other sensors and functions.

[0016] The hardware group includes a speaker 1206 as sound wave output means for outputting a sound wave generated from a set value of an inverse-phase waveform of predicted noise provided from an operation sound silencing processing unit 411 (see FIG. 5) described later. Further, the hardware group includes a sound collection microphone 1205 as sound collection means for collecting a combined sound of the noise during operation of the image forming apparatus 100 and the sound wave output from the speaker 1206. Further, the hardware group includes an in-machine temperature and humidity sensor 1207 for measuring the temperature and humidity inside the machine during operation, an out-of-machine temperature and humidity sensor 1208 for measuring the temperature and humidity outside the machine during operation, a display unit 1211 composed of a liquid crystal display input / output device, etc., an operation unit 1209 composed of a multi-touch sensor, etc.

[0017] The sound signal collected by the sound collection microphone 1205 is converted into digital data by, for example, an AD converter and stored as learning data in the HDD 1204. The speaker 1206 outputs a sound wave having a phase opposite to that of the predicted noise estimated by the estimation processing unit 405 described later. The temperature and humidity measured by the in-machine temperature and humidity sensor 1207 and the out-machine temperature and humidity sensor 1208 are acquired by the device state detection unit 407 (see FIG. 5) described later and stored as learning data in the HDD 1204. Further, the paper information such as the paper size, paper thickness, and paper type of the recording paper to be conveyed acquired from the operation unit 1209 and the general-purpose computer 103, and the printing mode are also stored as learning data in the HDD 1204. Further, the temperature and humidity measured by the in-machine temperature and humidity sensor 1207 and the out-machine temperature and humidity sensor 1208, the information such as the paper size, paper thickness, and paper type of the recording paper to be conveyed, and the printing mode are also used as input data input to the estimation processing unit 405 when the estimation processing unit 405 outputs the predicted noise.

[0018] The CPU 1201 is a control unit that comprehensively controls the image forming apparatus 100. The RAM 1202 is a system work memory for the operation of the CPU 1201 and an image memory for temporarily storing image data and the like. The network IF 1210 is connected to a LAN or the like and performs communication (transmission and reception) with external devices 300 such as the general-purpose computer 103, other computer terminals on the LAN, the machine learning server 102, and the data server 105. Further, it includes data communication with an external facsimile apparatus and a wireless communication function for wirelessly connecting to an external terminal.

[0019] The ROM 1203 stores programs and the like executed by the CPU 1201. The HDD 1204 stores system software, image data, software counter values, and the like. Note that other storage devices such as a solid state drive (SSD) may be provided instead of or in combination with the HDD.

[0020] The GPU1221 can perform calculations efficiently by processing more data in parallel. The GPU1221 may also be used for processing by the estimation processing unit 405 (see Figure 4), which predicts noise generated during the operation of the image forming apparatus, as described later. Note that the processing of the estimation processing unit 405, as described later, may also be configured to be performed solely by the CPU1201 or the GPU1221.

[0021] Figure 4 shows an example of the hardware configuration of the machine learning server 102. The machine learning server 102 comprises a CPU 1301, RAM 1302, ROM 1303, HDD 1304, a network interface 1310 (which serves as a transceiver), an I / O unit 1305, and a GPU 1306, all of which are interconnected via a system bus 1307.

[0022] The CPU 1301 provides various functions by reading and executing programs such as the OS (Operating System) and application software from the HDD 1304. The RAM 1302 is the system work memory used by the CPU 1301 when executing programs. The ROM 1303 stores programs for starting the BIOS (Basic Input Output System) and OS, as well as configuration files.

[0023] HDD1304 stores system software and other data. Alternatively, other storage devices such as SSDs may be provided instead of or in conjunction with the HDD. Network IF1310 is connected to a network and communicates (sends and receives data) with external devices such as the image forming apparatus 100.

[0024] The I / O unit 1305 is an interface for inputting and outputting information to and from the operation unit, which consists of a liquid crystal display input / output device equipped with a multi-touch sensor, etc. The operation unit is populated with predetermined information at a predetermined resolution, number of colors, etc., based on the screen information instructed by the program.

[0025] For example, a GUI (Graphical User Interface) screen is formed, and various windows and data necessary for operation are displayed on the control panel. Note that the control panel does not necessarily have to be present.

[0026] The GPU1306 can perform calculations efficiently by processing more data in parallel, so it is effective to use the GPU1306 when performing machine learning multiple times using a learning model such as deep learning. In this embodiment, the machine learning unit 414 (see Figure 5), which will be described later, uses the GPU1306 in addition to the CPU1301. Specifically, when executing a learning program that includes a learning model, the CPU1301 and GPU1306 work together to perform calculations and learn. Note that the processing of the machine learning unit 414 (see Figure 5), which will be described later, may also be performed by the CPU1301 or the GPU1306 alone.

[0027] The data server 105 and the general-purpose computer 103 can be implemented with the same hardware configuration as the machine learning server 102 described above. Furthermore, the data server 105 does not necessarily need to have an operating unit, similar to the machine learning server 102. Also, the machine learning server 102 and the data server 105 may be implemented on the same computer.

[0028] Furthermore, the machine learning server 102 and the data server 105 may be implemented by a single computer or by multiple computers. For example, the machine learning server 102 and the data server 105 may be implemented using cloud computing technology.

[0029] Figure 5 shows an example of the software configuration of the noise control system 1. The software for the noise control system 1 shown in Figure 5 is implemented by utilizing hardware resources and programs as shown in Figures 1, 3, and 4. The programs for implementing the software configuration shown in Figure 4 are stored in storage for each component, read into RAM, and executed by the CPU.

[0030] For example, in the image forming apparatus 100, the program for realizing the software configuration shown in Figure 5 is stored in the HDD 1204, read into the RAM 1202, and executed by the CPU 1201. In addition to the CPU 1201, the program may also be executed using the GPU 1221.

[0031] In the machine learning server 102, the program to implement the software configuration shown in Figure 4 is stored in the HDD 1304, read into RAM 1302, and executed by CPU 1301. In some cases, it may also be executed using GPU 1306 in addition to CPU 1301. Similarly, in the data server 105, the program to implement the software configuration shown in Figure 4 is stored in the HDD 1304, read into RAM 1302, and executed by CPU 1301.

[0032] The software configuration shown in Figure 5 is a pre-trained model, which is a noise prediction program trained using training data, that predicts the noise during operation of the image forming apparatus and generates an inverse phase waveform of the predicted noise. The generated inverse phase waveform is output from speaker 1206 (see Figure 2) during operation, and this is the software configuration of the noise control system 1 that realizes the function of reducing noise during operation.

[0033] As shown in Figure 5, the software of the image forming apparatus 100 includes a data storage unit 401 as a storage means for storing training data, a JOB control unit 403, an operation noise reduction processing unit 411 as an inverse phase waveform generation means, and an estimation processing unit 405 as a noise prediction means. It also includes an image reading unit 404, a counter unit 406, a device state detection unit 407, etc. Note that the image reading unit 404 is not necessary if the image forming apparatus 100 does not have an image reading device. The JOB control unit 403, the operation noise reduction processing unit 411 as an inverse phase waveform generation means, and the estimation processing unit 405 as a noise prediction means are held as part of the control program written to the control circuit 104 (see Figure 2).

[0034] The JOB control unit 403 has a functional role in executing the basic functions of the image forming apparatus 100, such as copying, faxing, and printing, based on user instructions, and in sending and receiving instructions and data between other software components in connection with the execution of these basic functions.

[0035] The device status detection unit 407 has a functional role in detecting the temperature and humidity inside and outside the image forming apparatus 100 from the internal temperature and humidity sensor 1207 and the external temperature and humidity sensor 1208, etc., when a print job is received. The detected device status is stored in the data storage unit 412 as training data or provided to the estimation processing unit 405 as input data.

[0036] The data storage unit 401 has a functional role in storing image data (print data), training data (described later), and trained models in the hardware configuration shown in Figure 3, relative to the RAM 1202 and HDD 1204. The following data is stored in the data storage unit 401 as training data. • Input data entered into the estimation processing unit 405 • Predicted noise data predicted by the estimation processing unit 405 • Noise data collected by the sound-collecting microphone 1205 during image formation operation while speaker 1206 is outputting an out-of-phase sound wave.

[0037] The input data entered into the estimation processing unit 405 is the machine operating conditions related to the fluctuations in noise generated during printing operations, and is acquired by the device status detection unit 407, etc., when a print job is received, etc. The input data is as follows: • Print mode (copy / print, monochrome / color, paper feed tray, paper feed speed (normal speed / half speed), etc.) • Temperature and humidity (outside and inside the aircraft) • Information about the paper to be fed (paper type, paper thickness, size, etc.) These are some examples, but the input data is not limited to those listed above.

[0038] The noise data collected by the sound-collecting microphone 1205 during the image formation operation is converted into digital data by an AD converter and stored in the data storage unit 401.

[0039] The training data is stored in the data storage unit 401 as a data table, linking the input data input to the estimation processing unit 405, the predicted noise data predicted by the estimation processing unit 405 from the input data, and the synthesized sound data of sound waves from the speaker 1206 collected by the sound collection microphone 1205 and the noise. The training data stored in this data storage unit 401 is transmitted to the data server 105 at a predetermined timing or in response to a request from the data server 105. In this embodiment, the training data is transmitted to the data server 105 each time training data is acquired.

[0040] The estimation processing unit 405 has the functional role of predicting the noise level during operation of the image forming apparatus based on the input data, using a trained model which is a noise prediction program that has been trained by machine learning on the machine learning server 102.

[0041] The noises generated during the operation (printing operation) of an image forming apparatus are mainly as follows: • Rotational noise from rotating parts such as motors, fans, and rollers • Sounds from the image formation process (motor, gear, etc. driving sounds, sliding sounds) • Noise from paper transport operation (noise from transport motors, feed motors, gears, etc.) • Rubbing noise between the paper transport roller and the paper transport guide

[0042] The estimation processing unit 405 predicts the noise level during operation (printing) when these sounds overlap, based on the input data described above (print mode, external and internal temperature and humidity, and paper information). This noise prediction is performed based on instructions from the JOB control unit 403. The estimation processing unit 405 is executed by the CPU 1201, GPU 1221, etc., as shown in Figure 2.

[0043] The prediction results made by the estimation processing unit 405 are sent from the JOB control unit 403 to the operation noise reduction processing unit 411, where they are reflected in the setting value of the inverse phase sound wave. Then, an inverse phase sound wave is generated from the reflected setting value and output from the speaker 1206, thereby reducing and eliminating noise during operation.

[0044] Furthermore, the prediction results predicted by the estimation processing unit 405 are transmitted to the JOB control unit 403, which displays a notification message on the operation unit 1209, thereby notifying the user that sound is being output from the speaker 1206. In addition, based on the display of the notification message, the user may be prompted to input their feedback on the noise control execution, such as the sound output from the speaker 1206 being too loud, and the volume of the sound emitted from the speaker 1206 may be adjusted based on this input data.

[0045] The image reading unit 404 has a functional role in controlling the execution of copy and scan functions based on instructions from the JOB control unit 403, including reading the original document with a reader and optically reading the recording paper with an in-line sensor within the device. The counter unit 406 has a functional role in recording and managing various counter values ​​(for example, the total number of printed pages) in the image forming apparatus 100.

[0046] The software for the data server 105 includes a data collection and provision unit 410 and a data storage unit 412. The data collection and provision unit 410 has the functional role of collecting and providing training data for training on the machine learning server 102. In the noise control system 1 of this embodiment, it has the functional role of collecting the above-mentioned training data from multiple image forming apparatuses 100 and providing it to the machine learning server 102. Furthermore, the data collection and provision unit 410 can be configured to receive training data from multiple image forming machines 100 connected to a network such as a LAN, and to transmit the training data collected from the multiple image forming machines to the machine learning server 102 upon request from the machine learning server 102. In this case, the machine learning unit 414's learning model can perform machine learning using its own training data and the training data from other image forming machines collected by the data collection and provision unit 410. This allows the learning model to be trained using a large amount of training data, thereby improving the accuracy of the learning model.

[0047] The data storage unit 412 has the functional role of recording and managing the training data collected by the data collection and provision unit 410. The functional roles of each software in the data server 105 are executed by the CPU of the data server 105.

[0048] The software for the machine learning server 102 includes a training data generation unit 413, a machine learning unit 414, and a data storage unit 415. The learning data generation unit 413 has a functional role in optimizing the received learning data to obtain the desired learning effect. For example, the learning data generation unit 413 optimizes the learning data by removing unnecessary data that would be noise from the learning data received from the data server 105, or by adjusting the format of the learning data for input into the learning model. It also synthesizes the predicted noise data included in the learning data with the synthesized sound collected by the sound collection microphone 1205 (a synthesized sound of noise during the operation of the image forming apparatus and sound waves output from the speaker 1206) to generate correct noise data. The learning data optimized by this learning data generation unit 413 is sent to the machine learning unit 414.

[0049] The data storage unit 415 has the functional role of temporarily recording the training data received from the data server 105, the training data optimized by the training data generation unit 413, and the training model in the machine learning unit 414 to the RAM 1302 and HDD 1304 shown in Figure 3.

[0050] The functional roles of the learning data generation unit 413 and the data storage unit 415 are performed by the CPU 1301, etc., as shown in Figure 3.

[0051] The machine learning unit 414 has a functional role in performing machine learning by using training data (teaching data) optimized by the training data generation unit 413 to obtain the desired learning effect, and by utilizing the hardware resources (GPU 1306, CPU 1301, etc.) shown in Figure 3 and the learning method using the learning model.

[0052] In this embodiment, the data server 105 and the machine learning server 102 that generates the learning model are located outside the image forming apparatus. However, the image forming apparatus may also be equipped with the functions of the data server 105 and the machine learning server 102. Figure 6 is a schematic diagram of the image forming apparatus 100 equipped with the functions of a data server 105 and a machine learning server 102, and Figure 7 is a diagram showing an example of a software configuration in which the image forming apparatus 100 is equipped with a data acquisition and provision unit, a machine learning unit, and a training data generation unit. The functions of the data server 105 and the machine learning server 102 are retained as part of the control program written to the control circuit 104. As shown in Figure 7, the image forming apparatus 100 includes a data collection and provision unit 410 that has the functional role of collecting training data, and a machine learning unit 414 that has the functional role of generating a trained model. The image forming apparatus 100 also includes a training data generation unit 413 that has the functional role of optimizing the training data provided to the machine learning unit 414. The data collection and provision unit 410 collects training data from image forming apparatuses of the same model and from a data server 105 via a network line, and the machine learning unit 414's training model uses its own training data and training data from other image forming apparatuses collected by the data collection and provision unit 410 to perform machine learning. This allows the training model to be trained using a large amount of training data, thereby improving the accuracy of the training model.

[0053] Figure 8 is a conceptual diagram showing an example of the input / output structure using a training model in the machine learning unit 414 of this embodiment, and Figure 9 is a diagram explaining machine learning in the machine learning unit 414. Here, a training model using a neural network is given as an example. As an example to explain the features of the noise control system 1 of this embodiment, the elements of the learning data (input layer) X involved in generating a learning model that uses device information as input to predict (output) noise generated during device operation are shown as X1 to X10. The input layer shown in Figure 8 is the input data described above.

[0054] Specific machine learning algorithms include neural networks, nearest neighbor methods, Naive Bayes, decision trees, and support vector machines. Deep learning, which uses neural networks to generate its own features and connection weights for learning, is another example. Appropriately, any of the above algorithms can be applied to this embodiment. Furthermore, the learning model may include an error detection unit and an update unit.

[0055] The machine learning unit 414 uses the optimized training data generated by the training data generation unit 413 as training data. The input data of this training data (input data input to the estimation processing unit 405) is defined as the elements (input layer) X of the training data. Furthermore, the expected value (T) is used to represent the correct noise data, which is obtained by combining the predicted noise data included in the training data with the synthesized sound data collected by the sound collection microphone 1205 (synthesized sound data of operating noise and sound output from the speaker).

[0056] Error detection involves obtaining the error between the output data (predicted noise data) Y, which is calculated by the neural network based on the input data X input to the input layer and output from the output layer, and the expected value (correct noise data) T, and calculating a loss L representing the error using a loss function. Based on the obtained loss L, the connection weight coefficients between nodes of the neural network are updated to minimize the loss (to bring the loss L closer to 0).

[0057] This update involves, for example, updating the connection weight coefficients between nodes in a neural network using backpropagation. Backpropagation is a method that adjusts the connection weight coefficients between nodes in each neural network to minimize the aforementioned error.

[0058] In this way, when input data X corresponding to the expected value (ground truth noise data) T is input to the learning model W, the weighting coefficients within the learning model W are adjusted so that the output data (predicted noise data) Y is as close as possible to the expected value (ground truth noise data) T, thereby obtaining a highly accurate learning model W. This process is called the learning process, and the learning model that has been adjusted through this learning process is called the "trained model".

[0059] In the noise control system 1 shown in Figures 1 to 5, where the machine learning unit 414 is located on the machine learning server 102, the "trained model" obtained in this way is sent to each image forming apparatus 100 via the network. When each image forming apparatus 100 receives the "trained model" from the machine learning server 102, it updates the "trained model" stored in the HDD 1204 with the received "trained model". For example, the "trained model" may be received from the machine learning server 102 when the power of the image forming apparatus 100 is turned ON.

[0060] On the other hand, as shown in Figures 6 and 7, if the machine learning unit 414 is provided in the image forming apparatus 100, the "trained model" stored in the HDD 1204 of the image forming apparatus is updated. The "trained model" may be updated each time the machine learning unit 414's training model is adjusted, or the "trained model" may be updated only when the accuracy of the training model has improved.

[0061] Figure 10 illustrates noise control during the operation of the image forming apparatus, and Figure 11 is a flowchart of the noise control. Figures 10 and 11 represent a configuration in which the image forming apparatus shown in Figures 6 and 7 is equipped with a machine learning unit 414. Furthermore, the left side of Figure 11 shows the print operation flowchart, and the right side shows the noise control flowchart. As shown in Figure 10, when a print job is received, machine operating conditions to be input to the estimation processing unit 405 are acquired (S1). Specifically, as described above, the internal temperature and humidity measured by the internal temperature and humidity sensor 1207 (see Figure 2), the external temperature and humidity measured by the external temperature and humidity sensor 1208 (see Figure 3), and the print mode and paper type information input to the operation unit 1209 are acquired as machine operating conditions to be input to the estimation processing unit 405. In addition, the internal temperature and humidity, external temperature and humidity, print mode, and paper type information acquired as machine operating conditions are stored in the data storage unit 401.

[0062] Next, the acquired machine operating conditions are input to the estimation processing unit 405 (trained model), which uses machine learning to predict the noise generated by the machine during printing (S2). Next, the predicted noise data output by the estimation processing unit 405 (trained model) is stored in the data storage unit 401, and based on the predicted noise data, the operation noise reduction processing unit 411 generates inverse-phase sound data of the predicted noise (S3). Next, the inverse-phase sound data generated by the operation noise reduction processing unit 411 is converted into an analog signal, and simultaneously with the start of machine operation (printing operation), an inverse-phase sound wave is output from the speaker 1206 (S4). As a result, the noise generated by the machine operation and the inverse-phase sound wave generated from the speaker cancel each other out, reducing the noise generated by the machine operation.

[0063] Simultaneously with the start of the above machine operation (printing operation), the sound-collecting microphone 1205 collects a composite sound of the machine noise and an out-of-phase sound wave output from the speaker 1206 (S5). The composite sound collected by the sound-collecting microphone 1205 is converted into digital data (composite sound data) by the AD converter. The composite sound data is then linked to the machine operation conditions as input data input to the estimation processing unit 405 (trained model) and the predicted noise data output from the estimation processing unit 405 (trained model), and stored in the data storage unit 401 as training data. This training data stored in the data storage unit 401 is sent to the machine learning unit 414 and processed into training data. Specifically, based on the composite sound data collected by the sound-collecting microphone 1205 and the predicted noise data, the correct noise data as the expected value T is generated. The machine learning unit 414 inputs the machine operating conditions as input data into the learning model, obtains the error between the predicted noise data output from the learning model and the correct noise data as the expected value T, and calculates a loss representing the error using a loss function. Based on the obtained loss, it updates the connection weight coefficients between nodes of the neural network and adjusts the learning model so that the loss is reduced (the loss L approaches 0). The learning model adjusted by the machine learning unit 414 is then used as a trained model to predict noise in the next noise control operation. As a result, in the next machine operation, noise during machine operation can be predicted with greater accuracy, and more effective noise reduction and silencing can be expected.

[0064] Furthermore, as shown in Figure 9, the machine learning unit 414 performs machine learning on the learning model using the training data received from the data server. This allows for training using a large amount of training data, thereby improving the accuracy of the learning model.

[0065] As mentioned above, the noise generated during machine operation consists of a combination of sounds, including rotational noise from rotating parts such as motors, fans, and rollers, driving and sliding noises from motors and gears, and friction noises between the paper transport and the paper transport path. While the rotational noise of motors and fans occurs at a constant speed during printing, the paper transport noise varies depending on the operating mode, paper type, and paper thickness. Furthermore, all of these sounds generated during machine operation change considerably due to individual differences between machines, the surrounding environment, and changes over time. Therefore, the noise reduction effect may be reduced with pre-prepared inverse-phase sound waves.

[0066] In contrast, this embodiment uses machine learning to predict the noise generated during machine operation and cancels out the noise during machine operation with sound waves that are in the opposite phase to the predicted noise. By using machine learning for noise control to suppress noise, it becomes possible to accurately predict noise that changes in various ways due to individual differences between machines, the surrounding environment, and changes over time, thereby exhibiting a more effective noise reduction and silencing function.

[0067] Alternatively, the estimation processing unit 405 (trained model) may be stored on an external device 300 (for example, a cloud system equipped with a machine learning server 102, a data server 105, etc.) that the image forming apparatus 100 communicates with via a network. Figure 12 is a schematic diagram showing the noise control system 1 when the estimation processing unit 405 (trained model) is installed in a cloud system 300, which is an external device located outside the image forming apparatus 100. In the noise control system 1 shown in Figure 12, the machine operating conditions of the image forming apparatus 100, such as internal and external temperature and humidity, printing mode, and paper type information, are transmitted as data to the cloud system 300, which acts as a noise prediction device. The image forming apparatus 100 receives predicted noise data predicted by the estimation processing unit 405 (trained model) on the cloud system 300 based on the received machine operating conditions. Then, based on the received predicted noise data, the operating noise reduction processing unit 411 (see Figure 5) of the image forming apparatus 100 generates inverse-phase sound data of the predicted noise, and outputs the generated inverse-phase sound from the speaker 1206 at the same time as the start of the machine operation (printing operation) of the image forming apparatus 100. With this, even an image forming apparatus without an AI function (trained model) can reduce noise based on predicted noise. Furthermore, in a configuration where the image forming apparatus 100 is equipped with an estimation processing unit 405 (a trained model), communication between the image forming apparatus 100 and the cloud system 300 at the time of printing becomes unnecessary, and the time from the user issuing a print command using the operation unit 1209 to the start of printing by the image forming apparatus 100 can be shortened.

[0068] Although preferred embodiments of the present invention have been described above, the present invention is not limited to these specific embodiments, and various modifications and changes are possible within the scope of the spirit of the present invention as described in the claims, unless otherwise specifically limited in the above description.

[0069] The above is just one example; each of the following embodiments produces its own unique effects. (Aspect 1) A noise control device comprising: noise prediction means such as an estimation processing unit 405 that predicts the noise to be generated based on operating conditions such as the machine operating conditions of an image forming apparatus 100; inverse phase waveform generation means such as an operating sound suppression processing unit 411 that generates an inverse phase waveform that is in the opposite phase to the predicted noise predicted by the noise prediction means; and sound wave output means such as a speaker 1206 that outputs sound waves of the inverse phase waveform generated by the inverse phase waveform generation means, wherein the noise prediction means predicts the noise to be generated by a noise prediction program such as a trained model that has been machine-learned using a plurality of training data. As mentioned above, the noise generated during the operation of an image forming apparatus is a combination of various sounds, including the noise from the drive mechanism, the sound of paper transport, and the rotation of the fan. Furthermore, the sound of paper transport is not constantly present during operation; it varies depending on the operating mode, paper type, paper thickness, etc. All of these sounds change in various ways due to individual differences between machines, the surrounding environment, and changes over time. Therefore, as described in Patent Document 1, the inverse phase waveform generated from predicted noise stored in a storage means has a low noise reduction effect. Therefore, in Embodiment 1, the noise generated is predicted based on a noise prediction program that has been machine-learned using multiple training data. By using machine learning to predict the noise of the image forming apparatus in this way, it is possible to accurately predict the noise during operation of the image forming apparatus, which is affected by various factors and where various sounds overlap. As a result, the noise of the image forming apparatus can be effectively reduced by the inverse phase waveform generated based on the predicted noise.

[0070] (Aspect 2) In Embodiment 1, the operating conditions, such as machine operating conditions, include printing conditions such as the printing mode, temperature and humidity during operation (internal and external temperature and humidity), and the type and thickness of paper being fed. According to this, the noise prediction program can accurately predict the noise generated during operation by predicting the noise based on factors such as printing conditions including the printing mode, operating temperature and humidity (inside and outside the machine), and the type and thickness of paper being fed through the image forming apparatus.

[0071] (Aspect 3) In embodiment 1 or 2, the system includes sound collection means such as a sound collection microphone 1205 that collects a composite sound of an inverse-phase waveform sound wave output by a sound wave output means such as a speaker 1206 and noise during the operation of the image forming apparatus, and storage means such as a data storage unit 401 that stores the composite sound collected by the sound collection means, operating conditions as machine operating conditions, and predicted noise predicted by a noise prediction means such as an estimation processing unit 405 as learning data. According to this, as described in the embodiment, machine learning can be performed using the training data stored in the storage means such as the data storage unit 401 to improve the accuracy of the learning model.

[0072] (Aspect 4) In embodiment 3, the noise prediction program, such as a trained model, is machine-learned using training data collected from multiple image forming apparatuses. According to this, as described in the embodiment, training can be performed using a large amount of training data, thereby improving the accuracy of the learning model.

[0073] (Aspect 5) In an image forming apparatus equipped with a noise control device, any noise control device from embodiment 1 to 4 was used as the noise control device. According to this, as described in the embodiment, the noise generated during the operation of the image forming apparatus can be effectively reduced.

[0074] (Aspect 6) In embodiment 5, the image forming apparatus has a transmitting and receiving unit such as a network IF 1210 that transmits and receives data with an external device 300. The transmitting and receiving unit transmits various data acquired from the image forming apparatus as training data to a data server 105 or the like provided in the external device 300, and also receives noise prediction programs such as trained models from a machine learning server 102 or the like provided in the external device 300. According to this, machine learning can be performed using an external device, which reduces the load on the image forming apparatus compared to performing machine learning within the image forming apparatus itself.

[0075] (Aspect 7) The noise control system 1 comprises an image forming apparatus 100 and an external device 300 such as a cloud system having a noise prediction means (estimation processing unit 405) that predicts the noise when the image forming apparatus 100 is operating. The image forming apparatus 100 transmits the operating conditions during image forming to the noise prediction means, the noise prediction means predicts the noise generated by a noise prediction program (trained model) that has been machine-learned using multiple training data, and transmits the predicted noise data to the image forming apparatus 100. The image forming apparatus 100 generates an inverse-phase waveform that is in the opposite phase to the predicted noise based on the received predicted noise data, and outputs sound waves of the generated inverse-phase waveform. According to this, even if the image forming apparatus does not have AI (Artificial Intelligence) capabilities, by utilizing AI functions provided in the cloud, it is possible to effectively reduce the noise of the image forming apparatus by generating an inverse phase waveform based on noise that accurately predicts the noise during operation of the image forming apparatus, which changes due to various factors. [Explanation of symbols]

[0076] 1: Image forming system 100: Image forming apparatus 102: Machine Learning Server 103: General-purpose computer 105: Data Server 300: External device (cloud system) 401: Data storage unit 403:JOB control section 404: Image reading unit 405: Estimation Processing Unit 407: Device status detection unit 410: Data Collection and Provision Department 411: Operation noise reduction processing unit 412: Data storage unit 413: Training Data Generation Unit 414: Machine Learning Department 415: Data storage unit 1205: Sound-collecting microphone 1206: Speaker 1207: In-flight temperature and humidity sensor 1208: External temperature and humidity sensor 1209:Operation unit 1210: Network Interface 1307: System bus 1310: Network Interface [Prior art documents] [Patent Documents]

[0077] [Patent Document 1] Japanese Patent Publication No. 2006-076206

Claims

1. A noise prediction means that predicts the noise during operation of the image forming apparatus based on the operating conditions of the image forming apparatus, The noise prediction means generates an inverse-phase waveform that is in the opposite phase to the predicted noise predicted by the noise prediction means, A noise control device comprising: a sound wave output means that outputs sound waves of the inverse phase waveform generated by the inverse phase waveform generation means, The noise prediction means is a noise control device characterized by predicting noise generated by a noise prediction program that has been machine-learned using multiple training data.

2. A noise control device according to claim 1, The noise control device is characterized in that the aforementioned operating conditions are printing conditions, temperature and humidity during operation, type of paper being fed, and paper thickness.

3. In the noise control device according to claim 1, The sound collection means collects a composite sound of the inverse-phase sound wave output by the sound wave output means and the noise during the operation of the image forming apparatus. A noise control device characterized by having a storage means for storing the synthesized sound collected by the sound collection means, the operating conditions, and the noise predicted by the noise prediction means as learning data.

4. In the noise control device according to claim 3, The noise control device is characterized in that the noise prediction program is machine-learned using the training data collected from a plurality of image forming apparatuses.

5. In an image forming apparatus equipped with a noise control device, An image forming apparatus characterized in that the noise control device described in claim 1 is used as the noise control device.

6. In the image forming apparatus according to claim 5, The image forming apparatus has a transmitting and receiving unit that transmits and receives data with an external device, The image forming apparatus is characterized in that the transmitting and receiving unit transmits the learning data to the external device and receives the noise prediction program generated by the external device from the external device.

7. Image forming apparatus and A noise control system comprising an external device having noise prediction means for predicting noise during the operation of the image forming apparatus, The image forming apparatus transmits the operating conditions during the image forming operation to the noise prediction means. The noise prediction means predicts the noise generated by a noise prediction program that has been machine-learned using training data, and transmits the predicted noise data to the image forming apparatus. The noise control system is characterized in that the image forming apparatus generates an inverse-phase waveform that is in the opposite phase to the predicted noise based on the received predicted noise data, and outputs sound waves of the generated inverse-phase waveform.

Images