Printer abnormality detection system, printer abnormality detection method and program

The printer abnormality sign detection system uses acoustic signal analysis and machine learning to accurately identify and predict failures in printer mechanisms, enhancing diagnostic precision and reducing user inconvenience.

JP7585746B2Active Publication Date: 2024-11-19TOPPAN HOLDINGS INC
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2020198772
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2020-11-30
Publication Date
2024-11-19
Estimated Expiration
2040-11-30

AI Technical Summary

Technical Problem

Existing printer failure detection methods lack accuracy in determining which mechanism within the printer has failed due to varying acoustic signal intensities and environmental factors, making it difficult to diagnose abnormalities in conveyance, transfer, and unloading processes.

Method used

A printer abnormality sign detection system that utilizes an acoustic signal acquisition unit, separation unit, signal processing unit, and machine learning model to generate three-dimensional time-series data, allowing for precise identification of abnormalities in each printing phase by analyzing spectral intensities and time-series data.

Benefits of technology

Enables accurate detection of abnormalities in printer mechanisms, predicting potential failures regardless of environmental conditions, and determining which mechanism is faulty, improving diagnostic precision and reducing user inconvenience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007585746000001
    Figure 0007585746000001
  • Figure 0007585746000002
    Figure 0007585746000002
  • Figure 0007585746000003
    Figure 0007585746000003
Patent Text Reader

Abstract

To provide a printer abnormality sign detection system that uses acoustic signals generated by transfer and carry-out mechanisms to detect abnormalities in each of transport, transfer, and carry-out of the printer, and to detect which of the respective mechanisms is abnormal or signs of an abnormality, even if a cause of the abnormality is unknown, regardless of an environment in which the printer is located.SOLUTION: A printer abnormality prediction detection system is a system for identifying and detecting an abnormality in each of printing phases of a printer, comprising: an acoustic signal acquisition unit for acquiring an acoustic signal of the printer; an acoustic signal separation unit for separating the acoustic signal into printing phase acoustic signals corresponding to each of the printing phases; a signal processing unit for extracting spectral intensity of each of a plurality of frequencies of the acoustic signal in a time series and generates three-dimensional time series data comprising each of the frequencies, the spectral intensity of the frequencies, and time; and an abnormality detection unit for detecting abnormalities in each of the printing phases by the three-dimensional time series data.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a printer abnormality sign detection system, a printer abnormality sign detection method, and a program for detecting a printer failure. [Background technology]

[0002] If a defect is found in a component mounted on a printer after the printer has been shipped, the printer must be recalled and, if a component defect is detected, the part must be replaced. However, it is inconvenient for users if a printer that is in use is recalled for repair. For this reason, it may be considered to install a failure detection program in the printer to detect signs of an upcoming failure and periodically diagnose the signs of a failure.

[0003] Meanwhile, there is a device that detects an abnormality in a gas turbine by acquiring an acoustic signal by an acoustic sensor and generating, for each frequency, two-dimensional time series data consisting of time and the spectral intensity of the frequency (for example, Patent Document 1). Then, the spectrum intensity of a specific frequency is compared with a preset threshold value, and the presence or absence of the above is determined based on whether or not the spectrum intensity exceeds the threshold value. By using the detection method of Patent Document 1, a printer failure can be detected by comparing the spectrum intensity of an acoustic signal generated by the printer when printing with a threshold value. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 08-177530 Summary of the Invention [Problem to be solved by the invention]

[0005] However, when detecting a failure in a device such as a printer, which is formed by combining multiple mechanisms such as transporting print media (paper, cards, etc.), transferring color (CMYK color components), and discharging, the accuracy of failure determination is low because it is not possible to determine whether the printer has failed or not. In other words, since the data used to detect abnormalities is two-dimensional time-series data, the acoustic signals generated when there is a malfunction in the transport, transfer, and discharge mechanisms are different, making it impossible to determine which mechanism has malfunctioned.

[0006] Furthermore, since the acoustic signal is detected in time-series data within a certain time range, it is difficult to include the acoustic signal during the operation of each of the multiple mechanisms in the printer in the detection range when it is not known when the mechanism will operate. Furthermore, since the spectral intensity of the acoustic signal is originally large, unlike a gas turbine for which the threshold can be fixed, the spectral intensity of the acoustic signal generated in the usage environment (due to changes in the vibration characteristics of the printer housing, etc.) also varies. For this reason, instead of fixing the threshold as in Patent Document 1, it is necessary to adjust the threshold to be compared with the acquired acoustic signal in accordance with the environment.

[0007] Moreover, the method of Patent Document 1 only determines whether or not the spectral intensity in the acoustic signal of the acquired time-series data is higher than a threshold value. For this reason, it is not possible to analyze information on the spectral intensity of the acoustic signal in the time domain before and after the time series data in order to make a judgment using information before and after the region with high spectral intensity. Furthermore, when the cause of the printer abnormality is unknown, it is not clear which frequency in the acoustic signal should be detected, or how high a threshold value should be set for the spectral intensity of the frequency to be detected.

[0008] The present invention has been made in consideration of the above-mentioned circumstances, and provides a printer abnormality detection system, printer abnormality detection method, and program that detects abnormalities in the printer's transport, transfer, and ejection mechanisms by acoustic signals generated by the operation of each mechanism after the printing operation has begun, and detects signs of which of the mechanisms is abnormal or may become abnormal (or malfunction), regardless of the environment in which the printer is placed, even if the cause of the abnormality (or malfunction) is unknown. [Means for solving the problem]

[0009] In order to solve the above-mentioned problems, the printer abnormality sign detection system of the present invention is a printer abnormality sign detection system that distinguishes and detects each abnormality in each printing phase of a printer, and includes an audio signal acquisition unit that acquires an audio signal of the printer, an audio signal separation unit that separates the audio signal into printing phase audio signals corresponding to each printing phase, and Printing Phase a signal processing unit that extracts spectral intensities of a plurality of frequencies of an acoustic signal in a time series and generates three-dimensional time series data consisting of each of the frequencies, the spectral intensities of the frequencies, and time; and an abnormality detection unit that detects an abnormality in each of the printing phases based on the three-dimensional time series data. the printing phase includes a transport process for transporting the printing medium, a printing process for printing ink, and a transport process for transporting the printing medium, the acoustic signal separation unit detects that the transport process has started based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a printing phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the printing phase acoustic signal of the printing process, and a transport area acoustic signal as the printing phase acoustic signal of the transport process, and each of the transport area acoustic signal, the print area acoustic signal, and the transport area acoustic signal are separated as the printing phase acoustic signal in a time range that includes all of the acoustic information in each of the transport, printing, and transport operations. It is characterized by:

[0010] The printer abnormality prediction detection system of the present invention comprises: A printer abnormality prediction detection system that distinguishes and detects abnormalities in each printing phase of a printer, the system comprising: an acoustic signal acquisition unit that acquires an acoustic signal of the printer; an acoustic signal separation unit that separates the acoustic signal into printing phase acoustic signals corresponding to each printing phase; a signal processing unit that extracts spectral intensities for each of a plurality of frequencies of the printing phase acoustic signal in a time series and generates three-dimensional time series data consisting of each of the frequencies, the spectral intensities of the frequencies, and time; and an abnormality detection unit that detects abnormalities in each of the printing phases using the three-dimensional time series data, the printing phases being a transport transport for transporting a print medium. The acoustic signal separation unit detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process and an unloading area acoustic signal as the print phase acoustic signal of the unloading process, and each of the transport area acoustic signal, the print area acoustic signal, and the unloading area acoustic signal is extracted so that other printing phase acoustic signals adjacent in time series overlap within a predetermined time range. It is characterized by:

[0011] The printer abnormality prediction detection system of the present invention comprises: The three-dimensional time series data is a spectrogram. It is characterized by:

[0012] The printer abnormality prediction detection system of the present invention comprises: The anomaly detection unit includes a machine learning model, and the machine learning model receives the three-dimensional time series data of the print phase acoustic signal as an input and outputs anomaly information indicating the degree of anomaly in each of the print phases. It is characterized by:

[0013] The present invention The printer abnormality sign detection method is a printer abnormality sign detection method that distinguishes and detects each abnormality in a printing phase of a printer, and includes an acoustic signal acquisition step in which an acoustic signal acquisition unit acquires an acoustic signal of the printer, an acoustic signal separation step in which an acoustic signal separation unit separates the acoustic signal into printing phase acoustic signals corresponding to each printing phase, a signal processing step in which a signal processing unit extracts spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in time series, a three-dimensional time series data generation step in which the signal processing unit generates three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time, and an abnormality detection step in which an abnormality detection unit detects each abnormality in each of the printing phases using the three-dimensional time series data. the printing phase includes a transport process for transporting a print medium, a printing process for printing ink, and a transport process for transporting the print medium, the acoustic signal separation unit detects that the transport process has started based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process, and a transport area acoustic signal as the print phase acoustic signal of the transport process, and each of the transport area acoustic signal, the print area acoustic signal, and the transport area acoustic signal are separated as the print phase acoustic signal in a time range that includes all of the acoustic information in each of the transport, printing, and transport operations. It is characterized by:

[0014] The present invention The printer abnormality sign detection method is a printer abnormality sign detection method that distinguishes and detects each abnormality in a printing phase of a printer, and includes an acoustic signal acquisition step in which an acoustic signal acquisition unit acquires an acoustic signal of the printer, an acoustic signal separation step in which an acoustic signal separation unit separates the acoustic signal into printing phase acoustic signals corresponding to each printing phase, a signal processing step in which a signal processing unit extracts spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in time series, a three-dimensional time series data generation step in which the signal processing unit generates three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time, and an abnormality detection step in which an abnormality detection unit detects each abnormality in each of the printing phases using the three-dimensional time series data. Contains the printing phase includes a transport process for transporting a print medium, a printing process for printing ink, and a carrying-out process for carrying out the print medium, the acoustic signal separation unit detects that the transport process has started based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process, and a carrying-out area acoustic signal as the print phase acoustic signal of the carrying-out process, and each of the transport area acoustic signal, the print area acoustic signal, and the carrying-out area acoustic signal is extracted such that other printing phase acoustic signals adjacent in time series overlap within a predetermined time range.It is characterized by:

[0015] The present invention The program causes a computer to function as a printer abnormality prediction detection system that distinguishes and detects abnormalities in each of the printing phases of a printer, and causes the computer to function as an acoustic signal acquisition means that acquires an acoustic signal of the printer, an acoustic signal separation means that separates the acoustic signal into printing phase acoustic signals corresponding to each of the printing phases, a signal processing means that extracts the spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in a time series and generates three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time, and an abnormality detection means that detects abnormalities in each of the printing phases using the three-dimensional time series data, and the printing phase is a period when a printing medium is transported. the acoustic signal separation means detects that the transport process has started based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a printing phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the printing phase acoustic signal of the printing process, and an unloading area acoustic signal as the printing phase acoustic signal of the unloading process, and each of the transport area acoustic signal, the print area acoustic signal, and the unloading area acoustic signal are separated as the printing phase acoustic signal in a time range that includes all of the acoustic information for each of the transport, printing, and unloading operations.

[0016] The program of the present invention is a program for causing a computer to function as a printer abnormality sign detection system that distinguishes and detects abnormalities in each printing phase of a printer, and the program causes the computer to include an acoustic signal acquisition means for acquiring an acoustic signal of the printer, an acoustic signal separation means for separating the acoustic signal into printing phase acoustic signals corresponding to each printing phase, Printing Phase a signal processing means for extracting a spectral intensity for each of a plurality of frequencies of an acoustic signal in a time series and generating three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time; and an anomaly detection means for detecting an anomaly in each of the printing phases using the three-dimensional time series data. the printing phase includes a transport process for transporting a printing medium, a printing process for printing ink, and a transport process for transporting the printing medium, and the acoustic signal separation means detects that the transport process has started based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a printing phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the printing phase acoustic signal of the printing process, and an unloading area acoustic signal as the printing phase acoustic signal of the unloading process, and each of the transport area acoustic signal, the print area acoustic signal, and the unloading area acoustic signal is extracted such that other printing phase acoustic signals adjacent in time series overlap within a predetermined time range. It is a program. Effect of the Invention

[0017] As described above, according to the present invention, a printer abnormality detection system, printer abnormality detection method, and program can be provided that detects abnormalities in each of the printer's transport, transfer, and ejection mechanisms by acoustic signals generated by the operation of each mechanism after the printing operation has begun, and detects signs of which of the mechanisms is abnormal or may become abnormal (or malfunction), regardless of the environment in which the printer is placed, even if the cause of the abnormality (or malfunction) is unknown. [Brief description of the drawings]

[0018] [Figure 1] 1 is a block diagram showing an example of the configuration of a printer abnormality sign detection system according to an embodiment of the present invention. [Diagram 2] 10A to 10C are waveform diagrams for explaining a process of separating an acoustic signal into each of the print phase acoustic signals. [Diagram 3] FIG. 2 is a diagram showing an example of a spectrogram used in the present embodiment. [Figure 4] 11 is a flowchart illustrating an example of the operation of a process for learning and generating a machine learning model using teacher data. [Diagram 5] 11 is a flowchart illustrating an example of the operation of a process for detecting a failure in a printer using a machine learning model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. 1 is a block diagram showing an example of the configuration of a printer abnormality sign detection system according to one embodiment of the present invention. The printer abnormality sign detection system 1 includes an acoustic signal acquisition unit 101, an acoustic signal separation unit 102, a signal processing unit 103, a machine learning model generation unit 104, an abnormality detection unit 105, a display unit 106, a result output unit 107, an acoustic signal storage unit 108, a machine learning model storage unit 109, and a determination result storage unit 110.

[0020] The acoustic signal acquisition unit 101 acquires acoustic signal data (hereinafter simply referred to as acoustic signal) as digital data of an acoustic signal detected by an acoustic sensor (not shown) provided in the printer or an acoustic sensor (not shown) placed as an external device near the printer. At this time, the acoustic signal acquisition unit 101 removes noise and unnecessary waveform data from the acoustic signal through a user editing process input from an input means (not shown, such as a keyboard) and generates waveform data suitable for generating three-dimensional time series data (such as a spectrogram, which will be described later).

[0021] Then, the acoustic signal acquisition unit 101 writes and stores the acquired and waveform-shaped acoustic signal data in the acoustic signal storage unit 108 together with the identification information of each printer and acquisition date and time information indicating the history. In addition, in the case of an acoustic signal used as training data when generating a machine learning model, fault information indicating the presence or absence of a fault for each printing phase in the printer that acquired the acoustic signal is added together with the acoustic signal data, and the data is written and stored in the acoustic signal storage unit 108.

[0022] The audio signal separator 102 separates the audio signal into print phase audio signals for each print phase of the printer. Fig. 2 is a waveform diagram for explaining the process of separating the acoustic signal into each printing phase acoustic signal. In the waveform diagram of Fig. 2, the horizontal axis indicates time, and the vertical axis indicates the amplitude level of the acoustic signal. The printer in this embodiment has a plurality of printing phases (printing phase #1, printing phase #2, printing phase #31, printing phase #4, printing phase #5, printing phase #6, and printing phase #7 described below) such as conveying the printing medium and printing or transferring ink to the printing medium, for example, as phases for printing on the printing medium.

[0023] In Fig. 2, the signal waveforms within the range of solid line frames (hereinafter, frames for setting the extraction time width) 501 to 507 are the printing phase sound signals of the printing phases and the signal waveforms to be extracted. Here, the signal waveform within frame 501 corresponds to printing phase #1, the signal waveform within frame 502 corresponds to printing phase #2, the signal waveform within frame 503 corresponds to printing phase #3, the signal waveform within frame 504 corresponds to printing phase #4, the signal waveform within frame 505 corresponds to printing phase #5, the signal waveform within frame 506 corresponds to printing phase #6, and the signal waveform within frame 507 corresponds to printing phase #7, and each of these corresponds to a printing phase. The printing phase sound signals during each of printing phase #2, printing phase #2, printing phase #2, printing phase #2, and printing phase #2 are, for example, print area sound signals. Also, for example, the printing phase sound signal during printing phase #1 is a transport area sound signal, and the printing phase sound signal during printing phase #7 is a discharge area sound signal. The above frames 501 to 507 each have the same time width. In this embodiment, for example, the time width of each of the frames 501 to 507 is in the range of 5 seconds. The height of the frame is set according to the amplitude width of the acoustic signal, and is set to a signal level width that includes the maximum amplitude in the acoustic signal.

[0024] The acoustic signal separation unit 102 reads out a first acoustic threshold value stored in advance in the acoustic signal storage unit 108. This first acoustic threshold value is a threshold value for determining the intensity of the signal level of an acoustic signal, which is used to detect the timing when the printing phase of card transport is started. The acoustic signal separation unit 102 then compares the discrete signal levels of the read acoustic signal with the first acoustic threshold value, and detects the time at which the signal level of the acoustic signal exceeds the first acoustic threshold value as the timing at which the printing phase of print medium transport starts.

[0025] When the acoustic signal separation unit 102 detects the timing when the print phase of the print medium transport starts, it moves the frame 501 back by a predetermined time length T501 (for example, 0.2 seconds in this embodiment) from the time of that timing. This time length T501 is set as a time length such that the acoustic signal when the print medium is actually being transported from the timing when the first acoustic threshold was exceeded in the printer that is the detection target in this embodiment is included within the frame (in this embodiment, moving back means moving (shifting) the frame in the opposite direction to the progression of time). By moving back the start time of the time width set by the time width of the frame 501 by the time length T501, the acoustic signal separation unit 102 adjusts the entire signal waveform to be the print phase acoustic signal (transport area acoustic signal) in the print medium transport so that it is included within the frame.

[0026] Here, the entire signal waveform of the acoustic signal generated during the print medium transport process is set to be included in frame 501, and the frame is set to extract the transport area acoustic signal as information for detecting the state of the print medium transport mechanism during the print medium transport process. The processing of each of frames 502 to 507 described below is performed in the same manner, and frames are set to extract the transfer area acoustic signal and the discharge area acoustic signal during the transfer process and the print medium discharge process, respectively.

[0027] Next, the acoustic signal separation unit 102 provisionally sets the frame 502 adjacent to the adjusted frame 501. Then, the acoustic signal separation unit 102 moves the frame 502 back by a predetermined time length T502 (for example, 1 second in this embodiment). This time length T502 is set as a time length to move back so that the acoustic signal when the print phase #2 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range of the time width of the frame 501 by a predetermined time length (so as to include the signal waveform in the print medium transport). By moving the start time of the time period of the frame 502 back by the time length T502, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (print area acoustic signal) in print phase #2 is included in the frame, and parts of the signal waveforms of the print phase acoustic signal (transport area acoustic signal) in the print medium transport and the print phase acoustic signal (print area acoustic signal) in print phase #3 are included.

[0028] Next, the acoustic signal separation unit 102 provisionally sets a frame 503 adjacent to the adjusted frame 502. Then, the acoustic signal separation unit 102 moves the frame 503 back by a predetermined time length T503 (for example, 1 second in this embodiment). This time length T503 is set as a time length to move back so that the acoustic signal when the print phase #3 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range of the time width of the frame 502 by a predetermined time length (so as to include the signal waveform in print phase #2). By moving the start time of the time period of the frame 503 back by the time length T503, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (print area acoustic signal) in print phase #3 is included in the frame, and parts of the signal waveforms of the print phase acoustic signal (print area acoustic signal) in print phase #2 and the print phase acoustic signal (print area acoustic signal) in print phase #4 are included.

[0029] Next, the acoustic signal separation unit 102 provisionally sets a frame 504 adjacent to the adjusted frame 503. Then, the acoustic signal separation unit 102 moves the frame 504 back by a predetermined time length T504 (for example, 1.2 seconds in this embodiment). This time length T504 is set as a time length to move back so that the acoustic signal when the print phase #4 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range in the time width of the frame 503 by a predetermined time length (so as to include the signal waveform in the print phase #3). By moving the start time of the time period of the frame 504 back by the time length T504, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (print area acoustic signal) in print phase #4 is included in the frame, and parts of the signal waveforms of the print phase acoustic signal (print area acoustic signal) in print phase #3 and the print phase acoustic signal (print area acoustic signal) in print phase #5 are included.

[0030] Next, the acoustic signal separation unit 102 provisionally sets a frame 505 contiguous to the adjusted frame 504. Then, the acoustic signal separation unit 102 moves the frame 505 back by a predetermined time length T505 (for example, 1.2 seconds in this embodiment). This time length T505 is set as a time length to move back so that the acoustic signal when the print phase #5 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range of the time width of the frame 504 by a predetermined time length (so as to include the signal waveform in the print phase #4). By moving the start time of the time period of the frame 505 back by the time length T505, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (print area acoustic signal) in the print phase #5 is included in the frame, and parts of the signal waveforms of the print phase acoustic signal (print area acoustic signal) in the print phase #4 and the print phase acoustic signal (print transfer area acoustic signal) in the secondary transfer are included.

[0031] Next, the acoustic signal separation unit 102 provisionally sets a frame 506 adjacent to the adjusted frame 505. Then, the acoustic signal separation unit 102 moves the frame 506 back by a predetermined time length T506 (for example, 1 second in this embodiment). This time length T506 is set as a time length to move back so that the acoustic signal when the print phase #6 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range of the time width of the frame 505 by a predetermined time length (so as to include the signal waveform in the print phase #5). By moving the start time of the time period of the frame 506 back by the time length T506, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (print area acoustic signal) in print phase #6 is included in the frame, and parts of the signal waveforms of the print phase acoustic signal (print area acoustic signal) in print phase #5 and the print phase acoustic signal (discharge area acoustic signal) in discharging the print medium are included in the frame.

[0032] Next, the acoustic signal separation unit 102 provisionally sets a frame 507 adjacent to the adjusted frame 506. Then, the acoustic signal separation unit 102 moves the frame 507 back by a predetermined time length T507 (for example, 0.5 seconds in this embodiment). This time length T507 is set as a time length to move back so that the acoustic signal when the print phase #7 is actually being performed is included in the frame in the printer to be detected in this embodiment, and the frame has a time period that overlaps with the time range of the time width of the frame 506 by a predetermined time length (so as to include the signal waveform in the print phase #6). By moving the start time of the time period of the frame 507 back by the time length T507, the acoustic signal separation unit 102 adjusts the frame so that the entire signal waveform of the print phase acoustic signal (discharge area acoustic signal) in the print phase #7 is included in the frame, and a part of the signal waveform of the print phase acoustic signal (print area acoustic signal) in the print phase #6 is included in the frame.

[0033] As described above, the acoustic signal separation unit 102 detects the timing when the printing phase of the printing medium transport begins, sets frame 501 at that timing, adjusts the time band of frame 501 with time length T501, and sequentially sets frames 502, 503, 504, 505, 506, and 607, and adjusts the time band of frames 502, 503, 504, 505, 506, and 607 with time lengths T502, T503, T504, T505, T506, and T507, respectively, thereby separating (extracting) the printing phase acoustic signals of printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7 from the acoustic signal by extracting the acoustic signals within the above-mentioned frames.

[0034] In addition, the acoustic signal separation unit 102 may detect the timing when the printing phase of print medium transport has started, set frame 501 at that timing, and then the user may adjust the duration T501 of the time period to be extracted (cut out) by frame 501, set each of frames 502 to 507, and adjust the start times of the frame durations by the durations T502 to T507 by moving the frame with a mouse or the like while observing the display screen of display unit 106.

[0035] The signal processing unit 103 performs signal processing on each of the print phase acoustic signals extracted by the acoustic signal separation unit 102 using frames, and generates a spectrogram (three-dimensional time series data). Here, the signal processing unit 103 uses a short-time Fourier transform or a group of digital band-pass filters to extract the spectral intensity for each frequency of each printing phase acoustic signal in time series, and converts the printing phase acoustic signal into a spectrogram. A spectrogram is a type of three-dimensional time-series data graph in which the vertical axis represents frequency and the horizontal axis represents time, and the signal strength of the frequency at each time (spectral strength, i.e., sound intensity, dB) is represented by color (e.g., multiple color gradations).

[0036] 3 is a diagram showing an example of a spectrogram used in this embodiment, in which the vertical axis represents frequency, the horizontal axis represents time, and the color of each coordinate point of frequency and time represents signal intensity. As a result, the spectrogram contains not only the signal strength of a specific frequency, but also information on the signal strength of multiple frequencies before and after that time, making it three-dimensional time series data that encompasses more information on sounds generated during the printing phase than conventional examples. Therefore, compared to a case where the signal strength of a specific frequency is compared with a threshold value to determine the presence or absence of a fault, it is possible to analyze the signal strength before and after the time for each frequency, and the correlation with the signal strength at other frequencies. Therefore, it is possible to obtain not only the presence or absence of a fault based on whether or not the threshold value is exceeded, but also information predicting the occurrence of a future fault based on the fault probability (a type of abnormality information, which indicates the degree of the fault state, and the probability indicated in numerical steps from a fault state to a normal state), such as whether the state is closer to a fault or closer to normal.

[0037] Returning to FIG. 1, the machine learning model generation unit 104 receives the spectrogram as input and generates a machine learning model that outputs a failure probability indicating a state of failure (failure that leads to failure) of each mechanism in the printing phase. The machine learning model generation unit 104 learns from the print phase acoustic signal obtained from a normal printer and the print phase signal obtained from a printer in an abnormal state (a broken state or a state that is about to develop into a breakdown) as training data, and generates a machine learning model. Then, the machine learning model generation unit 104 writes the generated machine learning model into the machine learning model storage unit 109 for storage. Here, examples of the machine learning model that is used include deep learning, neural networks, genetic programming, inductive logic programming, and support vector machines.

[0038] Furthermore, the machine learning model generation unit 104 generates a machine learning model by training it using predetermined teacher data, as described below. At this time, print phase acoustic signals for each print phase are extracted from the acoustic signals acquired from a normal printer, and are used as training data for the machine learning model. Similarly, from the acoustic signals acquired from a printer in an abnormal state, acoustic signals for each printing phase are extracted and used as training data for the machine learning model. The machine learning model used in this embodiment takes, for example, the spectrogram of each printing phase as input, and outputs the type of printing phase and the failure probability of that printing phase. The output is, for example, a numerical value between "1" and "0."

[0039] In this embodiment, for example, the inputs to the machine learning model are spectrograms of seven printing phases: printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7. The outputs of the machine learning model include 14 outputs, for example, output values ​​indicating the probability (printing phase probability) of printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7, and output values ​​indicating the probability of failure (failure probability) of each printing phase. In addition, in this embodiment, a printer with seven printing phases is described as an example, but if there are n (an integer greater than or equal to 2) printing phases, there will be 2n outputs, with n outputs indicating the printing phase probability for each printing phase and n outputs indicating the failure probabilities for each printing phase.

[0040] The following process is carried out when learning the print phase acoustic signal during normal operation, which is obtained from an acoustic signal obtained from a printer that is operating normally and has no malfunctions, as teaching data. For example, a spectrogram generated from the print phase acoustic signal in print phase #1 is input as training data to a machine learning model, and the machine learning model is trained to output a print phase probability of print phase #1 as a type of print phase closer to "1" and a card transport failure probability closer to "0." The print phase probabilities of print phases other than print phase #1 are trained to be closer to "0." Similarly, the machine learning model is trained using spectrograms for each of the other printing phases, namely, print phase #2, print phase #3, print phase #4, print phase #5, print phase #6, and print phase #7, obtained from a printer that is operating normally and has no malfunctions.

[0041] On the other hand, when a print phase acoustic signal during abnormal operation, obtained from an acoustic signal acquired from a broken printer performing an abnormal operation, is used as training data for learning, the following process is carried out. For example, a spectrogram generated from the print phase acoustic signal in print phase #1 is input to a machine learning model, and the machine learning model is trained to output a print phase probability of print phase #1 as a print phase type close to "1" and a failure probability of print phase #1 close to "1". In this case, an acoustic signal obtained from a printer in which print phase #1 is faulty (or in an abnormal state) is used. The failure probability of print phases other than print phase #1 is trained to be close to "0". The print phase probability of print phases other than print phase #1 is trained to be close to "0".

[0042] Similarly, the machine learning model is trained using spectrograms of the other printing phases, namely, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7, acquired from the faulty printer. In this case, acoustic signals from a printer in which each printing phase is faulty are used. The failure probability of the printing phases other than the faulty printing phase is trained to be close to "0". As described above, a machine learning model is generated by repeatedly learning from spectrograms obtained from multiple normal printers and spectrograms obtained from multiple printers with failed printing phase functions as training data. For example, if the spectrogram of printing phase #2 is input into the machine learning model after training described above, the printing phase probabilities output for printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7 are "0.09", "0.70", "0.00", "0.00", "0.00", and "0.21", respectively, with the printing phase probability of printing phase #2 being output as "0.70", the highest.

[0043] In addition, when new acoustic signals are obtained from a normal or broken printer, rather than from the printer of the training data stored in the acoustic signal storage unit 108, the machine learning model adds these acoustic signals as new training data to the training data group, and re-learns (updates) the machine learning model, thereby improving the accuracy of the failure probability of the machine learning model.

[0044] Then, the abnormality detection unit 105 uses the machine learning model to perform an operation to detect the presence or absence of a malfunction in each printing phase of the printer, as shown below. That is, the anomaly detection unit 105 reads out the machine learning model from the machine learning model storage unit 109, and judges the input spectrogram of each printing phase. In this case, by inputting the spectrograms of each of the printer's printing phases into the machine learning model, the printing phase probability for each printing phase, printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7, is output.

[0045] As a result, when, for example, the probability that the printing phase of printing phase #1 is closest to "1" is output and the failure probability of printing phase #7 is a number closest to "1", the printer abnormality prediction detection system 1 (result output unit 107 described later) displays (notifies) information indicating that "an abnormal sound is occurring in printing phase #1" on the display unit 106. The anomaly detection unit 105 writes and stores the failure probability indicating an abnormal state of each printing phase, which is output from the machine learning model as a result of the judgment, in the judgment result storage unit 110 for each printer. The display unit 106 is, for example, a device equipped with a liquid crystal panel for displaying images. The result output unit 107 reads out the determination result from the determination result storage unit 110, and notifies the user of the result by displaying an image on the display unit 106 or by voice from a speaker (not shown).

[0046] As described above, according to this embodiment, a machine learning model is used that takes as input a spectrogram of the printing phase acoustic signal of each printing phase extracted from the acoustic signal, and outputs a failure probability for each printing phase. Therefore, even if the cause of the abnormality is unknown, it is possible to estimate the presence or absence of an abnormality in each printing phase. Furthermore, rather than simply detecting that the printer is broken, it is possible to determine whether or not a mechanism in any printing phase of the printer is broken, or to estimate an abnormal state, such as determining whether or not the mechanism is in an abnormal state.

[0047] Furthermore, according to this embodiment, each of the spectrograms (printing phase acoustic signals) to be judged includes the time regions before and after each printing phase on the time axis, and therefore the relationship between the printing phases is also used to estimate the printing phase probability and failure probability of each printing phase, thereby improving the estimation accuracy of the printing phase probability and failure probability using information on other factors.

[0048] Furthermore, according to this embodiment, the machine learning model estimates from the spectrogram which printing phase the printing phase acoustic signal belongs to as the printing phase probability, so there is no need for the user to be aware of which printing phase the printing phase acoustic signal belongs to and perform a judgment process, thereby eliminating the need for the user to go through the trouble of confirming the printing phase. Furthermore, according to this embodiment, since a machine learning model is used for the judgment, there is no need to set a specific threshold value to judge whether or not there is an abnormality in the printing phase, and therefore, when judging whether or not there is an abnormality in the printing phase, there is no need to perform processing such as changing the threshold value depending on the surrounding environment (increasing the degree of freedom).

[0049] Furthermore, according to this embodiment, the machine learning model determines the time change in signal strength of each frequency component of the acoustic signal in three dimensions, and uses the multidimensional information of the acoustic signal to estimate and output which one or more printing phases in the printer are in an abnormal or failed state. Therefore, based on the interrelationships of the multidimensional information, it is possible to obtain the probability of an abnormal state, including failure, with high accuracy. In other words, the numerical value of the failure probability (for example, a value of 0.5 or more indicates an abnormal state that is close to failure) can be used to predict the occurrence of a failure in the mechanism of each printing phase of the printer.

[0050] FIG. 4 is a flowchart showing an example of the operation of a process of learning and generating a machine learning model using training data. Step S101: The acoustic signal acquisition unit 101 receives input of each piece of acoustic signal data (digital data) as teacher data supplied from an external device. The acoustic signal acquisition unit 101 sequentially writes and stores the acquired acoustic signal data and pass / fail information indicating that the printer is normal or that the printer is broken and which printing phase is abnormal in the acoustic signal storage unit 108. As a result, the acoustic signal storage unit 108 accumulates a plurality of pieces of acoustic signal data that serve as teacher data for training the machine learning model.

[0051] Step S102: The acoustic signal separation unit 102 selects and reads out data of any one of the acoustic signals from the acoustic signal storage unit 108 .

[0052] Step S103: The acoustic signal separation unit 102 determines whether or not the data length of the acoustic signal data is equal to or greater than a prescribed length that is set in order to perform separation processing efficiently. At this time, if the data length of the acoustic signal data is equal to or greater than the specified length, the acoustic signal separation unit 102 advances the process to step S104, whereas if the data length of the acoustic signal data is less than the specified length, the acoustic signal separation unit 102 advances the process to step S105.

[0053] Step S104: The audio signal separation unit 102 performs pre-processing for separation processing by removing unnecessary data such as noise and silence at both ends of the audio signal. The above-mentioned pre-processing for removing unnecessary data from the acoustic signal may be configured so that the user visually recognizes and removes unnecessary data while observing the waveform of the acoustic signal displayed on the display unit 106.

[0054] Step S105: The acoustic signal separation unit 102 compares the signal level (intensity value) of the acoustic signal from the start time of the acoustic signal with a first acoustic threshold (predetermined peak value) to detect the time during which the signal level exceeds the first acoustic threshold.

[0055] Step S106: The sound signal separation unit 102 then sets the time when the signal level exceeds the first sound threshold as the timing when print medium transport in the printing phase starts (sound information from the print medium transport mechanism starts), that is, as the start timing.

[0056] Step S107: As a result, the audio signal separation unit 102 aligns the end (start time) of a frame (frame 501) of a predetermined time width with the start timing, and positions frame 501 so that it includes the entire sound of the print medium transport mechanism. The portion of the acoustic signal contained within this frame 501 is extracted as the print phase acoustic signal corresponding to the print phase of print phase #1 print medium delivery.

[0057] Thereafter, as already explained, the acoustic signal separation unit 102 uses frame 501 as a reference and sequentially separates (extracts) the printing phase acoustic signals of printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, and printing phase #7 using frames 502 to 507. Then, the audio signal separator 102 writes and stores each of the separated print phase audio signals in the audio signal storage unit 108 in association with the audio signal into which the print phase signal has been separated.

[0058] Step S108: The signal processing unit 103 reads, from the audio signal storage unit 108, the print phase audio signal extracted from the audio signal to be processed. Then, the signal processing unit 103 converts each of the read printing phase acoustic signals into a spectrogram, associates it with the acoustic signal to be processed, and writes and stores it in the acoustic signal storage unit 108 .

[0059] Step S109: The acoustic signal separation unit 102 determines whether or not the process of separating all of the acoustic signals stored in the acoustic signal storage unit 108 has been completed. At this time, if the processing for separating all of the acoustic signals stored in the acoustic signal storage unit 108 is completed, the acoustic signal separation unit 102 advances the process to step S110. On the other hand, if the process of separating all of the acoustic signals stored in the acoustic signal storage unit 108 has not been completed, the acoustic signal separation unit 102 returns the process to step S102.

[0060] Step S110: The machine learning model generation unit 104 sequentially reads out, from the machine learning model storage unit 109, a spectrogram of the printing phase acoustic signal for each acoustic signal, as well as the type of printing phase corresponding to the spectrogram and pass / fail information indicating the presence or absence of an abnormality in each printing phase, as training data, and causes the machine learning model to learn. In other words, the teaching data includes a spectrogram for each printing phase, and as a label, information indicating whether the printing phase of the spectrogram is printing phase #1, printing phase #2, printing phase #3, printing phase #4, printing phase #5, printing phase #6, or printing phase #7, and whether or not there is an abnormality in the printing phase mechanism ("1" if there is, "0" if there is not).

[0061] Then, the machine learning model generation unit 104 inputs the spectrogram into the machine learning model and trains the machine learning model so that the printing phase probability, which is the output of the printing phase corresponding to the input spectrogram, is close to "1", the printing phase probabilities of the other printing phases are close to "0", and the failure probability is close to "1" if there is an abnormality in the printing phase and close to "0" if there is no abnormality.

[0062] Step S111: After the machine learning model has finished learning using all the training data stored in the acoustic signal storage unit 108, the machine learning model generation unit 104 determines whether the number of learning cycles is equal to or greater than a predetermined number of cycles, i.e., whether the learning has been repeated a predetermined number of times. At this time, if the machine learning model generation unit 104 has repeated the learning of the machine learning model a predetermined number of times, it proceeds to step S112, whereas if the number of times the machine learning model has been learned has not reached the predetermined number of times, it returns to step S110. In addition, the machine learning model generation unit 104 may be configured to repeat the learning process a predetermined number of times, but to repeatedly use all of the training data stored in the acoustic signal storage unit 108 until the printing phase probability is equal to or greater than a predetermined value and the numerical values ​​of the failure probabilities are equal to or greater than a predetermined value corresponding to the presence or absence of each failure.

[0063] Step S112: The machine learning model generation unit 104 repeats the learning of the above-mentioned machine learning model a predetermined number of times using all of the training data stored in the acoustic signal storage unit 108, thereby completing the generation of the machine learning model.

[0064] Step S113: Then, the machine learning model generation unit 104 writes the generated machine learning model into the machine learning model storage unit 109 for storage.

[0065] FIG. 5 is a flowchart showing an example of the operation of a process for detecting a failure in a printer using a machine learning model. Step S201: The acoustic signal acquisition unit 101 writes and stores in the acoustic signal storage unit 108 the acoustic signal data of the target printer for which the presence or absence of a fault is to be detected, the data being supplied from an external device.

[0066] Step S202: The acoustic signal separation unit 102 selects and reads out from the acoustic signal storage unit 108 acoustic signal data that is the subject of determination of the presence or absence of an abnormality. Here, the processes in steps S202 to S208 are similar to the processes in steps S102 to S108 already described, respectively, and therefore will not be described.

[0067] Step S209: The anomaly detection unit 105 reads out from the machine learning model storage unit 109 each of the spectrograms of the print phase acoustic signals obtained from the acoustic signals of the target printer for determining the presence or absence of an anomaly. In addition, the anomaly detection unit 105 reads out the machine learning model from the machine learning model storage unit 109.

[0068] Step S210: The anomaly detection unit 105 inputs each of the read spectrograms of the printing phases to a machine learning model. As a result, the anomaly detection unit 105 estimates and outputs the print phase probability of each spectrogram and the failure rate of that print phase using a machine learning model. In other words, when a spectrogram of a printing phase in which the printing phase is unknown and the presence or absence of an abnormality is unknown is input, the machine learning model outputs the printing phase probability for each printing phase as a number between "1" and "0", and outputs the failure probability for each printing phase as a number between "1" and "0".

[0069] Step S211: The abnormality detection unit 105 writes and stores the print phase probability of each of the print phases and the failure probability of each of the print phases obtained corresponding to the input spectrogram in the judgment result storage unit 110 as a judgment result. In addition, the abnormality detection unit 105 may be configured to determine, for example, the printing phase with the highest printing phase probability corresponding to a spectrogram as the printing phase of that spectrogram, and to determine the failure probability corresponding to that printing phase as the abnormal state of that printing phase. In addition, multiple predetermined thresholds may be set for the failure probability, and a failure may be determined if a first threshold is exceeded, and a failure may be determined if a second threshold (< first threshold) is exceeded but is less than the first threshold, indicating that there is an abnormality that is a precursor to a failure.

[0070] 1 in the present invention may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed to extract print phase acoustic signals from the acoustic signals of the printer, and a machine learning model may be used to detect the presence or absence of an abnormality in each print phase or to detect signs of failure. Note that the term "computer system" here includes hardware such as the OS and peripheral devices. Additionally, "computer system" also includes WWW systems equipped with a homepage providing environment (or display environment). Additionally, "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computer systems. Furthermore, "computer-readable recording medium" also includes devices that hold programs for a certain period of time, such as volatile memory (RAM) within computer systems that act as servers or clients when a program is transmitted via a network such as the Internet or a communication line such as a telephone line.

[0071] The above program may also be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The above program may also be for realizing part of the above-mentioned functions. Furthermore, it may be a so-called difference file (difference program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system. [Explanation of symbols]

[0072] 1. Printer abnormality detection system 101...acoustic signal acquisition unit 102...Acoustic signal separation section 103...signal processing unit 104…Machine learning model generation unit 105…Abnormality detection unit 106…Display section 107...Result output section 108...acoustic signal storage unit 109…Machine learning model memory section 110…Judgment result storage unit

Claims

1. A printer abnormality prediction and detection system that distinguishes and detects abnormalities in each printing phase of a printer, an acoustic signal acquisition unit for acquiring an acoustic signal of the printer; an acoustic signal separation unit that separates the acoustic signal into print phase acoustic signals corresponding to each of the print phases; a signal processing unit that extracts a spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in a time series manner and generates three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection unit that detects an anomaly in each of the printing phases based on the three-dimensional time series data; Equipped with the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separation unit detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process and a discharge area acoustic signal as the print phase acoustic signal of the discharge process; The transport area acoustic signal, the print area acoustic signal, and the discharge area acoustic signal are each separated as the print phase acoustic signal in a time range that includes all of the acoustic information in each of the transport, printing, and discharge operations. A printer abnormality prediction detection system comprising:

2. A printer abnormality prediction and detection system that distinguishes and detects abnormalities in each printing phase of a printer, an acoustic signal acquisition unit for acquiring an acoustic signal of the printer; an acoustic signal separation unit that separates the acoustic signal into print phase acoustic signals corresponding to each of the print phases; a signal processing unit that extracts a spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in a time series manner and generates three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection unit that detects an anomaly in each of the printing phases based on the three-dimensional time series data; Equipped with the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separation unit detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process and a discharge area acoustic signal as the print phase acoustic signal of the discharge process; The transport area acoustic signal, the print area acoustic signal, and the discharge area acoustic signal are each extracted so that other print phase acoustic signals adjacent in time series overlap with each other within a predetermined time range. A printer abnormality prediction detection system comprising:

3. The three-dimensional time series data is a spectrogram.

3. The printer abnormality prediction detection system according to claim 1 or 2.

4. The anomaly detection unit includes a machine learning model, The machine learning model receives the three-dimensional time series data of the printing phase acoustic signal as an input, and outputs anomaly information indicating the degree of anomaly in each of the printing phases.

4. The printer abnormality prediction detection system according to claim 1, wherein the printer abnormality prediction detection system detects an abnormality in the printer.

5. A printer abnormality sign detection method for determining and detecting abnormalities in each printing phase of a printer, an acoustic signal acquisition step in which an acoustic signal acquisition unit acquires an acoustic signal of the printer; an acoustic signal separation step in which an acoustic signal separation unit separates the acoustic signal into print phase acoustic signals corresponding to each of the print phases; a signal processing step in which a signal processing unit extracts a time series of spectral intensities of the printing phase acoustic signal for each of a plurality of frequencies; a three-dimensional time series data generating process in which the signal processing unit generates three-dimensional time series data including each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection step in which an anomaly detection unit detects an anomaly in each of the printing phases using the three-dimensional time series data; Including, the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separation unit detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process and a discharge area acoustic signal as the print phase acoustic signal of the discharge process; The transport area acoustic signal, the printing area acoustic signal, and the discharge area acoustic signal are each separated as the printing phase acoustic signal in a time range that includes all of the acoustic information in each of the transport, printing, and discharge operations. A printer abnormality sign detection method comprising:

6. A printer abnormality sign detection method for determining and detecting abnormalities in each printing phase of a printer, an acoustic signal acquisition step in which an acoustic signal acquisition unit acquires an acoustic signal of the printer; an acoustic signal separation step in which an acoustic signal separation unit separates the acoustic signal into print phase acoustic signals corresponding to each of the print phases; a signal processing step in which a signal processing unit extracts a time series of spectral intensities of the printing phase acoustic signal for each of a plurality of frequencies; a three-dimensional time series data generating process in which the signal processing unit generates three-dimensional time series data including each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection step in which an anomaly detection unit detects an anomaly in each of the printing phases using the three-dimensional time series data; Including, the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separation unit detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process and a discharge area acoustic signal as the print phase acoustic signal of the discharge process; The transport area acoustic signal, the print area acoustic signal, and the discharge area acoustic signal are each extracted so that other print phase acoustic signals adjacent in time series overlap with each other within a predetermined time range. A printer abnormality sign detection method comprising:

7. A program that causes a computer to function as a printer abnormality prediction detection system that distinguishes and detects abnormalities in each printing phase of a printer, The computer, an acoustic signal acquiring means for acquiring an acoustic signal of the printer; an acoustic signal separating means for separating the acoustic signal into print phase acoustic signals corresponding to the respective print phases; a signal processing means for extracting a spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in a time series manner and generating three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection means for detecting an anomaly in each of the printing phases based on the three-dimensional time series data; Functioning as a the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separating means detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process, and a discharge area acoustic signal as the print phase acoustic signal of the discharge process, The transport area acoustic signal, the print area acoustic signal, and the discharge area acoustic signal are each separated as the print phase acoustic signal in a time range that includes all of the acoustic information in each of the transport, printing, and discharge operations. program.

8. A program that causes a computer to function as a printer abnormality prediction detection system that distinguishes and detects abnormalities in each printing phase of a printer, The computer, an acoustic signal acquiring means for acquiring an acoustic signal of the printer; an acoustic signal separating means for separating the acoustic signal into print phase acoustic signals corresponding to the respective print phases; a signal processing means for extracting a spectral intensity for each of a plurality of frequencies of the printing phase acoustic signal in a time series manner and generating three-dimensional time series data consisting of each of the frequencies, the spectral intensity of the frequency, and time; an anomaly detection means for detecting an anomaly in each of the printing phases based on the three-dimensional time series data; Functioning as a the printing phase includes a conveying process for conveying a print medium, a printing process for printing ink, and a conveying process for conveying the print medium, the acoustic signal separating means detects the start of the transport process based on the intensity of the acoustic signal, extracts a transport area acoustic signal as a print phase acoustic signal corresponding to the start of the transport process, and then sequentially separates a print area acoustic signal as the print phase acoustic signal of the printing process, and a discharge area acoustic signal as the print phase acoustic signal of the discharge process, The transport area acoustic signal, the print area acoustic signal, and the discharge area acoustic signal are each extracted so that other print phase acoustic signals adjacent in time series overlap with each other within a predetermined time range. program.

Citation Information

Patent Citations

  • Abnormality detecting device for gas turbine

    JP1996177530A

  • Sound identification condition setting support device and sound identification condition setting support method

    JP2015175770A

  • Diagnostic device, diagnostic method, program and diagnostic system

    JP2017120622A

  • Printing device, printer diagnostic system, and printer diagnostic method

    JP2017208803A

  • Image forming apparatus and image forming method

    JP2020106773A