JP2016146020A - Data analysis system and analysis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data analysis system capable of performing a failure prediction at real time on large number of devices all over the world at higher frequency while avoiding convergence of data load to a reception section.SOLUTION: The analysis system which includes one or more information processors determines network devices as the objects from which data required for a second analysis is collected from plural network devices based on a result of a first analysis. The analysis system controls to collect the data required for the second analysis from the determined network devices at a higher frequency than the frequency of data collection in the first analysis. The analysis system executes the second analysis by using the data collected from the determined network devices.SELECTED DRAWING: Figure 8

Description

  The present invention relates to a reduction in system load in a failure prediction system that predicts a failure of an image forming apparatus in advance.

  In recent years, movements to actively analyze and utilize big data have become active. Also in the field of image forming apparatuses, it is considered to create a failure prediction model from sensor information and failure history from a large number of devices and perform failure prediction of devices in the world in real time. This system periodically collects sensor data from the image forming apparatus and applies a failure prediction model to the collected data. If an abnormal value is found as a result of the application, a serviceman is requested to replace or clean the parts requiring maintenance of the image forming apparatus. As a result, it is possible to prevent a failure from occurring at that location.

  As a prior art for performing more efficient maintenance in such a system, there is, for example, Patent Document 1. In Patent Document 1, when an abnormality is detected in any device with respect to a device failure prediction, another device installed in the same area as the device is specified. Then, for each of the identified devices, a criterion raising determination process is performed to determine whether there is an abnormality with a stricter determination criterion. Since it is often the case that a similar device abnormality is found in the same region, efficient maintenance work can be performed by the technique of Patent Document 1.

JP 2007-49349 A

  In the technique described in Patent Document 1, when data is acquired from a large number of image forming apparatuses existing all over the world, a load is concentrated on the data receiving unit. Furthermore, the frequency of failure prediction increases as the frequency of data reception from the device increases, but the load on the receiving unit further increases accordingly.

  In order to solve the above problems, the present invention is an analysis system that includes one or more information processing apparatuses and analyzes data collected from a plurality of network devices, and the network device is based on the result of the first analysis. Determining means for determining a network device to collect data necessary for the second analysis, and from the determined network device at a frequency higher than the frequency of data collection in the first analysis. Control means for controlling the data necessary for the second analysis to be collected, and execution means for performing the second analysis using the data collected from the determined network device Provide an analysis system.

  In the present invention, since data is normally received from the device at a low frequency, the load on the data receiving unit can be suppressed. If an abnormality is determined in failure prediction using low-frequency data, the device is instructed to send data at high frequency, and failure prediction is performed using high-frequency data. Can be suppressed.

Overall configuration of data utilization system Server computer hardware configuration Data storage software configuration Data analyzer software configuration Device management software configuration Hardware configuration of image forming apparatus Image forming device software configuration Low load failure prediction flowchart High-frequency transmission history DB status management flowchart High-frequency transmission history DB state management flowchart considering serviceman regular dispatch schedule Overall sequence of data utilization system Data collector software configuration Analyst terminal software configuration

  The best mode for carrying out the present invention will be described below with reference to the drawings. Note that “S” in the flowchart represents a step.

<Example 1>
FIG. 1 is a diagram illustrating an overall configuration of a failure prediction system according to Embodiment 1 of the present invention. In FIG. 1, a device management apparatus 111, a data storage apparatus 112, a data analysis apparatus 113, a plurality of network devices (121 to 123), a data collection apparatus 114, and an analyst terminal 115 are connected via networks 101 to 105. Note that the data storage device 112, the data analysis device 113, and the data collection device 114 in FIG. 1 may be included in the analysis system and each may be realized as a virtual machine. These virtual machines are executed on one or more information processing apparatuses.

  Although the network device is described as an image forming apparatus in this specification, the present invention can also be applied to a digital medical device, a network camera, a robot, an in-vehicle terminal, or an air conditioner. Although the number of network devices is described as three, the number is not limited to three. The networks 101 to 105 are, for example, a LAN such as the Internet, a WAN, a telephone line, a dedicated digital line, an ATM, a frame relay line, a cable television line, a data broadcasting wireless line, or the like. Or it is what is called a communication network implement | achieved by these combination. The networks 101 to 105 only need to be able to transmit and receive data. In this specification, the network 104 is assumed to be the Internet, and the networks 101 to 103 and 105 are assumed to be corporate networks or service provider networks, but may be other networks.

  The device management apparatus 111, the data storage apparatus 112, and the data analysis apparatus 113 are executed by a server computer (information processing apparatus). The device management apparatus 111 can set the sensor information transmission frequency and the sensor information transmission period to the image forming apparatuses 121 to 123, and can instruct sensor information transmission or stop sensor information transmission. Sensor information will be described later with reference to FIG. Further, the image forming apparatuses 121 to 123 can acquire sensor values installed in the image forming apparatuses 121 to 123 and transmit them to the data storage apparatus 112. The data storage device 112 can store the information received from the image forming devices 121 to 123 and transmit data necessary for failure prediction to the data analysis device 113. The data includes at least one of sensor data, counter information, and log information.

  The data analysis device 113 acquires data necessary for failure prediction from the data storage device 112, applies a failure prediction model previously installed in the data analysis device 113, and sends the determination result to the device management device 111. You can be notified. The failure prediction model in the data analysis device 113 is installed in advance by the analyst through the analyzer terminal 115. The data collection device 114 can acquire data necessary for creating a failure prediction model by an analyst from the data storage device 112 and store the data so that it can be accessed from the analyst terminal 115 at any time.

  The analyst terminal 115 is operated by an analyst. The analyst creates a failure prediction model of the image forming apparatus by machine learning or a simple mathematical formula using the sensor data and the failure history for a certain period in the analyst terminal 115. Sensor data and failure history can be acquired from the data collection device 114. The created failure prediction model can be implemented in the data analysis device 113.

  Hereinafter, each function of the server described in this specification may be realized by a single server or a single virtual server, or may be realized by a plurality of servers or a plurality of virtual servers. Alternatively, a plurality of virtual servers may be executed on a single server.

  FIG. 11 represents the overall sequence of the present invention. The present invention provides a “model creation phase” for creating a failure prediction model, a “low-precision failure prediction phase” for acquiring sensor data at a low frequency and predicting failure, and a “high” for acquiring sensor data at a high frequency and performing failure prediction. It consists of three phases of “accuracy failure prediction phase”.

  In the model creation phase, the analyst first notifies the data analyzer 113 of a sensor data transmission instruction request M1101 through the analyzer terminal 115. Upon receiving the transmission instruction request M1101, the data analysis apparatus 113 notifies the device management apparatus 111 of a sensor data transmission instruction request M1102. Upon receiving the transmission instruction request M1102, the device management apparatus 111 notifies the image forming apparatuses 121 to 123 of a sensor data transmission instruction M1103. The image forming apparatuses 121 to 123 transmit the sensor data to the data storage apparatus 112 in accordance with the sensor data transmission frequency and transmission period designated by the transmission instruction M1103 (M1104). Further, when a failure occurs in the image forming apparatuses 121 to 123, a failure notification M 1105 is notified from the image forming apparatuses 121 to 123 to the device management apparatus 111. Further, at an arbitrary timing, the data collection device 114 notifies the data storage device 112 of sensor data acquisition M1106 to acquire sensor data, and notifies the device management device 111 of failure history acquisition M1107 to cause a failure. Get history.

  When the sensor data and failure history necessary for model creation are accumulated in the data collection device 114, the analyst notifies the data collection device 114 through the analyst terminal 115 of the sensor data acquisition M1108 to obtain the sensor data. To do. Then, the analyst notifies the data collection device 114 of the failure history acquisition M1109 through the analyzer terminal 115, and acquires the failure history. In the analyst terminal 115, the analyst creates a failure prediction model from the acquired sensor data and failure history by machine learning or a simple mathematical formula (M1110). Then, the analyst installs the created model in the data analysis apparatus 113 through the analyst terminal 115 (M1111).

  If one or more models are mounted in the data analysis apparatus 113 through the above process, the low-accuracy failure prediction phase can be entered. In the low-precision failure prediction phase, the analyst first notifies the data analysis device 113 of the failure prediction start M1121 through the analyzer terminal 115. Once the failure prediction start M1121 is notified, the processing in the low accuracy failure prediction phase or the high accuracy failure prediction phase is repeatedly executed. In response to the notification of the failure prediction start M1121, the data analysis device 113 notifies the device management device 111 of the low frequency transmission instruction request M1122 in order to acquire sensor data necessary for failure prediction. The “low frequency transmission instruction” is a transmission instruction in which a transmission frequency that can be specified as an argument is set low. Upon receiving the notification of the low frequency transmission instruction request M1122, the device management apparatus 111 notifies the image forming apparatuses 121 to 123 of the low frequency transmission instruction M1123. The image forming apparatuses 121 to 123 transmit the sensor data to the data storage apparatus 112 in accordance with the sensor data transmission frequency and transmission period designated by the low frequency transmission instruction M1123 (M1124).

  The data analysis device 113 periodically monitors the data storage device 112. As a result, if sensor data that can be applied to the failure prediction model of the data analysis device 113 is stored, the sensor data acquisition M1125 is notified to the data storage device 112 to acquire the sensor data. The data analysis device 113 performs failure prediction M1126 using the acquired sensor data. As a result, when it is determined that there is an abnormality, the process shifts to alt [when it is determined that there is an abnormality], which is a high-frequency failure prediction phase.

  In the high-frequency failure prediction phase, the data analysis device 113 notifies the device management device 111 of a high-frequency transmission instruction request M1141 in order to acquire sensor data necessary for applying the high-precision failure prediction model. The “high frequency transmission instruction” is a transmission instruction with a high transmission frequency that can be specified as an argument. Upon receiving the notification of the high-frequency transmission instruction request M1141, the device management apparatus 111 notifies the image forming apparatuses 121 to 123 of the high-frequency transmission instruction M1142. The image forming apparatuses 121 to 123 transmit sensor data to the data storage apparatus 112 in accordance with the sensor data transmission frequency and transmission period designated by the transmission instruction M1142 (M1143).

  The data analysis device 113 periodically monitors the data storage device 112. As a result, if sensor data sufficient to apply the failure prediction model of the data analysis device 113 is accumulated, the sensor data acquisition M1144 is notified to the data storage device 112 to acquire the sensor data. The data analysis device 113 performs failure prediction M1145 using the acquired sensor data. As a result, when it is determined that there is an abnormality, the process moves to alt [when it is determined that there is an abnormality] that is a high-frequency failure prediction phase, and the data analysis apparatus 113 notifies the device management apparatus 111 of a sensor error notification M1146. To do.

  FIG. 2 is a hardware configuration diagram of the device management apparatus 111, the data storage apparatus 112, the data analysis apparatus 113, the data collection apparatus 114, and the analyst terminal 115 according to the first embodiment. In FIG. 2, a central processing unit (hereinafter “CPU”) 202 controls the entire apparatus. Specifically, the CPU 202 executes an application program, OS, and the like stored in a hard disc drive (hereinafter “HDD”) 205. Further, the CPU 202 performs control to temporarily store information, files, and the like necessary for program execution in a random access memory (hereinafter “RAM”) 203. A RAM 203 is a temporary storage unit, and functions as a main memory, a work area, and the like of the CPU 202. A Read Only Memory (hereinafter “ROM”) 204 is a storage unit and stores various data such as a basic I / O program. The HDD 205 is one of external storage means and functions as a large-capacity memory, and stores application programs such as a Web browser, a service server group program, an OS, and related programs.

  A display 208 is a display unit that displays commands and the like input from the keyboard 207. An interface 209 is an external device interface and connects a printer, a USB device, and a peripheral device. The keyboard 207 is an instruction input unit. The system bus 201 controls the flow of data in the apparatus. A network interface card (hereinafter “NIC”) 206 exchanges data with external devices via the networks 101 to 105.

  Note that the configuration of the computer is an example thereof, and is not limited to the configuration example of FIG. For example, the storage destination of data and programs can be changed by the ROM 204, the RAM 203, the HDD 205, etc. according to the characteristics. In addition, when the CPU 202 executes a process based on a program stored in the HDD 205, the software configuration as shown in FIG.

  FIG. 3 is a software configuration diagram of the data storage device 112 according to the first embodiment. The data storage device 112 includes a data reception unit 301, a data acquisition unit 302, and a sensor data management DB 303. The data receiving unit 301 has a function of receiving sensor information transmitted from the image forming apparatuses 121 to 123 and storing it in the sensor data management DB 303. The “sensor information” received here refers to the acquisition date and time of sensor data and the sensor data of each sensor. For example, it corresponds to one row in Table A below.

  The data acquisition unit 302 has a function of receiving a data acquisition request from the data analysis device 113 or the data collection device 114 and returning the value of sensor data managed by the sensor data management DB 303 as a return value. The sensor data management DB 303 is a database that manages sensor data received from the image forming apparatus. Data managed by the sensor data management DB 303 is sensor data as shown in Table A below, for example.

  Table A is an example of sensor data when the values of sensors M, N, and O attached to the image forming apparatus are collected every hour. The “data acquisition date” column in Table A is the date and time when the sensor data was acquired from the image forming apparatus. The “model” column represents the model name of the image forming apparatus. The “model number” column represents the model number of the image forming apparatus, and is a unique value for each image forming apparatus. The “Sensor M”, “Sensor N”, and “Sensor O” columns represent sensor values of “data acquisition date / time” of the image forming apparatus uniquely identified by “model number”. The sensor value is, for example, temperature in the image forming apparatus, humidity, voltage applied to components, and the like.

  FIG. 4 is a software configuration diagram of the data analysis apparatus 113 according to the first embodiment. The data analysis apparatus 113 includes a failure prediction unit 401, an external data acquisition unit 402, a sensor data transmission instruction unit 403, and a sensor error notification unit 404, as shown in the upper part of the figure. Further, as shown in the lower part, a failure prediction model list management DB 405, a high frequency transmission history DB 406, a high frequency transmission history management unit 407, and a sensor data transmission instruction request reception unit 408 are included.

  The failure prediction unit 401 has a function of applying a failure prediction model to data acquired from the data storage device 112. The failure prediction model can be installed in the failure prediction unit 401 through the analyzer terminal 115 by the analyst. When a certain failure prediction model is mounted in the failure prediction unit 401, the failure prediction model list management DB 405 is also updated and a new failure prediction model is registered. The failure prediction process will be described at the end of the description of FIG. The failure prediction unit 401 also controls the start and stop of failure prediction according to instructions from the analyst.

  The external data acquisition unit 402 has a function of acquiring sensor data stored in the data storage device 112. The sensor data transmission instruction unit 403 has a function of notifying the device management apparatus 111 to transmit sensor data of a specified frequency and period to a specified (target) image forming apparatus (for example, 121). Depending on how the argument is set, an instruction to stop transmission is also possible.

The sensor error notification unit 404 has a function of notifying the device management apparatus 111 that the designated image forming apparatus (for example, 121) has been determined to be a sensor abnormality as a result of failure prediction. The failure prediction model list management DB 405 is a database that manages a list of failure prediction models for making failure predictions. The data managed by the failure prediction model list management DB 405 is, for example, data as shown in Table B below, and corresponds to the failure prediction model installed in the failure prediction unit 401 by the analyst.

  The “model name” column in Table B is a name of a failure prediction model that is unique for each failure prediction model. The “model” column is the same as the “model” column in Table A. The “input sensor” column is a type of sensor data necessary for applying the model, and the type of sensor data written in this column becomes an input to the failure prediction model. The “input frequency” column is an input frequency of sensor data input to the failure prediction model. The “input period” column is an input period of sensor data input to the failure prediction model. The “model group” column represents a model group to which the failure prediction model belongs. The “high frequency level” column is used to manage which failure prediction model is to be applied next when high frequency output occurs in the model group. The “Notification content” column indicates the content to be notified to the device management device when an abnormality is determined by applying a certain failure prediction model. The “high frequency transmission instruction” or “sensor error” (failure notification) Either. The “high frequency transmission instruction” is an instruction for requesting the device management apparatus 111 to transmit sensor data at a frequency higher than that of the device management apparatus 111 in order to apply the model of the same group and the next level. The “continue application” column indicates whether or not to continue applying the model even after determining an abnormality by applying a certain failure prediction model.

  In Table B, Model A, B, and C belong to the same model group A, the high frequency levels are 0, 1, and 2 in order, and Model A, which is a model of level 0, is applied first. As a result, when abnormality is determined, Model B which is level 1 of the same model group A is applied next. In this case, since “continuation of application” of Model A is TRUE, application of Model A is continued even after that. When abnormality is determined as a result of applying Model B, Model C, which is a level 2 model of the same model group A, is applied next. When the abnormality is determined in Model C, since the “notification content” of Model C is a sensor error, the failure prediction unit 401 requests the sensor error notification unit 404 to notify the device management apparatus 111 of a sensor error.

  The high-frequency transmission history DB 406 is a database that manages a history of notifying the device management apparatus 111 of a high-frequency transmission instruction. Data managed by the high-frequency transmission management DB 406 is data as shown in Table C below, for example. As a result of the failure prediction unit 401 performing failure prediction, an abnormality is indicated, and a new row is updated when the sensor data transmission instruction unit 403 is notified of a high-frequency transmission instruction.

  The “model number” column in Table C is the same as the “model number” column in Table A. The “applied high frequency model” column represents the model name of the high frequency model applied by the device. The “high frequency model” refers to a model having a high frequency level of 1 or more, and the low frequency model refers to a model having a high frequency level of 0. The “status” column represents the status of the image forming apparatus corresponding to the model number, and takes a status of “high frequency transmission completed” or “high frequency transmission in progress”. The “high frequency start” column represents the time when the device receives a high frequency transmission instruction and starts high frequency transmission. The “scheduled high frequency stop” column represents the date and time when the device is scheduled to stop high frequency transmission.

  The high-frequency transmission history management unit 407 has a function of monitoring data in Table C managed by the high-frequency transmission history DB 406. The high-frequency transmission history is constantly monitored at a fixed period, and if the current date and time exceeds the scheduled high-frequency stop date and time of the high-frequency transmission history, a high-frequency transmission stop instruction is sent to the image forming device of the model number corresponding to the history. The data transmission instruction unit 403 is notified. Furthermore, the state of the history is changed from “high frequency transmission” to “high frequency transmission completed”. The sensor data transmission instruction request receiving unit 408 has a function of receiving a transmission instruction request from outside and notifying the sensor data transmission instruction unit 403 of the received transmission instruction.

  A processing flow of low load failure prediction in the failure prediction unit 401 will be described with reference to the flowchart of FIG. This processing flow corresponds to a message notified by the data analysis device 113 in the low accuracy failure prediction phase and the high accuracy failure prediction phase of the sequence diagram of FIG.

  In step S801, the failure prediction unit 401 acquires necessary sensor data from the data storage device. Acquisition of sensor data is performed by the failure prediction unit 401 inquiring of the external data acquisition unit 402. Next, the failure prediction unit 401 determines a failure prediction model to be applied in S802. A model to be applied to the sensor data acquired in the previous step is determined from the failure prediction model list managed by the failure prediction model list management DB 405.

  The application model is determined in S811 to S813. First, in S811, the “applicable high frequency model” registered in the high frequency transmission history DB 406 is determined as the application model. A line of the model number corresponding to the sensor data acquired in S801 is searched from a list such as Table C included in the high-frequency transmission history DB 406, and if it is found, the “applicable high-frequency model” is determined as the applicable model. To do. In step S812, “continue application” determines the TRUE model as the application model. For all the application models determined in S811, all models having the same “model group” as the application model and “high frequency level” lower than the application model are extracted, and “continue application” is TRUE among the extracted models. All models are determined as applicable models. In step S813, a model having a high frequency level of 0 is determined as an application model. All applied models determined in S811 and S812 are confirmed, and all model groups to which no model has been applied are extracted. Then, all models having a high frequency level of 0 for the extracted model group are determined as application models.

  In the loop from S803 to S809, all the models determined in S802 are applied, and the respective processes are performed. First, in S804, the failure prediction model is applied to the sensor data acquired in S801. In S805, the process branches depending on the determination result of model application, and if an abnormal value is indicated, the process proceeds to S806. If the normal value is indicated or the sensor information necessary for the model is insufficient, the process proceeds to S809.

  In S806, the “notification content” item of the application model is confirmed, and the process branches depending on the content. If the notification content item is “high frequency transmission instruction”, the process proceeds to S807, and if it is “sensor error”, the process proceeds to S808. In S807, a high-frequency transmission instruction is issued. The failure prediction unit 401 requests the sensor data transmission instruction unit 403 for the high frequency transmission instruction, and the sensor data transmission instruction unit 403 directly requests the device management apparatus 111 for the high frequency transmission instruction. At this time, the failure prediction unit 401 notifies the sensor data transmission instruction unit 403 of a transmission instruction. Here, the frequency specified in the transmission instruction uses the value of “input frequency” in Table B corresponding to the application model, and the period specified in the transmission instruction uses the value of “input period” in Table B corresponding to the application model. Use.

  In S808, a sensor error is notified and a high-frequency transmission stop is instructed. The sensor error notification is performed by the failure prediction unit 401 inquiring of the sensor error notification unit 404 so that the sensor error notification unit 404 directly notifies the device management apparatus 111 of the sensor error. The failure prediction unit 401 requests the sensor data transmission instruction unit 403 to instruct the high frequency transmission stop, and the sensor data transmission instruction unit directly requests the device management apparatus 111 to stop the high frequency transmission.

  FIG. 5 is a software configuration diagram of the device management apparatus 111 according to the first embodiment. The device management apparatus 111 includes a transmission instruction reception unit 501, a sensor error notification reception unit 502, a device transmission instruction unit 503, a serviceman instruction unit 504, a failure notification reception unit 505, a failure history acquisition unit 506, and a failure history management DB 507. The

  The transmission instruction receiving unit 501 has a function of receiving a high-frequency transmission instruction from the data analysis apparatus 113. It is possible to request the device transmission instruction unit 503 to interpret the content of the instruction from the data analysis apparatus 113 and to instruct the designated image forming apparatus to perform high-frequency transmission. The sensor error notification receiving unit 502 has a function of receiving a sensor error notification from the data analysis device 113. The content of the notification from the data analyzer 113 can be interpreted, the content of the sensor error can be notified to the serviceman instruction unit 504, and the device transmission instruction unit 503 can be notified of a high frequency transmission stop instruction.

  The device transmission instruction unit 503 has a function of receiving an instruction from the transmission instruction receiving unit 501 or the sensor error notification receiving unit 502 and directly notifying the specified instruction to the specified image forming apparatus (for example, 121). The instruction content to the image forming apparatus includes a sensor data transmission frequency, a transmission period, and a sensor data transmission stop instruction. The serviceman instructing unit 504 has a function of receiving the notification from the sensor error notification receiving unit 502 and directly notifying the serviceman of the content of the sensor error.

  The failure notification receiving unit 505 has a function of receiving a failure history transmitted from the image forming apparatuses 121 to 123 and registering the failure history in the failure history management DB 507. The failure history acquisition unit 506 has a function of receiving a failure history acquisition request from the outside and returning failure history data registered in the failure history management DB 507. Data managed by the failure history management DB 507 is, for example, data as shown in Table D below. The failure history from the image forming apparatuses 121 to 123 acquired by the failure notification receiving unit 505 is managed.

  The “failure occurrence date and time” column in Table D represents the date and time when the failure occurred. The “model” column has the same meaning as the “model” column in Table B. The “model number” column has the same meaning as the “model number” column in Table C. The “failure code” column is a code number corresponding to the type of failure that has occurred one to one.

  FIG. 6 is a hardware configuration diagram of the image forming apparatuses 121 to 123 according to the first embodiment. A controller 601 controls the entire image forming apparatus. Specifically, the controller 601 acquires sensor data from the sensor group 603 and stores the value in the HDD 604. Here, the sensor group 603 represents a set of various sensors in the image forming apparatus. For example, a temperature sensor or a voltage sensor is included. The HDD 604 can store all sensor data obtained from the sensor group 603. A network interface card (NIC) 602 exchanges data with external devices via the networks 101 to 105.

  FIG. 7 is a software configuration diagram of the image forming apparatuses 121 to 123 according to the first embodiment. The image forming apparatuses 121 to 123 include a transmission instruction reception unit 701, a sensor data transmission unit 702, an in-device sensor data management DB 703, and a failure notification unit 704. The transmission instruction receiving unit 701 has a function of receiving a sensor data transmission instruction from the device management apparatus 111. The instruction content from the device management apparatus 111 includes a sensor data transmission frequency, a sensor data transmission period, and a sensor data transmission stop instruction, and issues an instruction to the sensor data transmission unit 702 according to the instruction content.

  The sensor data transmission unit 702 has a function of directly transmitting sensor data to the data storage device 112. In response to an instruction from the transmission instruction receiving unit 701, data held in the in-device sensor data management DB 703 can be directly transmitted to the data storage device 112. Data managed by the in-device sensor data management DB 703 is, for example, data as shown in Table E below. Sensor data acquired by the sensor group 603 in the image forming apparatus is recorded. The failure notification unit 704 has a function of detecting various failures occurring in the image forming apparatuses 121 to 123 and notifying the device management apparatus 111 of them.

  The “data acquisition date” column and “sensor M” to “sensor O” columns in Table E are the same as the “data acquisition date and time” columns and “sensor M” to “sensor O” columns in Table A.

  FIG. 12 is a software configuration diagram of the data collection device 114 according to the first embodiment. The data collection device 114 includes an external data acquisition unit 1201, a data acquisition request reception unit 1202, a sensor data management DB 1203, and a failure history management DB 1204. The external data acquisition unit 1201 has a function of acquiring sensor data or a failure history from the data storage device 112 or the data management device 111. When the sensor data is acquired, the acquired data is registered in the sensor data management DB 1203, and when the failure history is acquired, the acquired data is registered in the failure history management DB 1204.

  The data acquisition request receiving unit 1202 has a function of receiving a data acquisition request from the analyst terminal 115 and returning data registered in the sensor data management DB 1203 or the failure history management DB 1204. The sensor data management DB 1203 manages the data acquisition date and time and the value of each sensor, similar to Table E. The failure history management DB 1204 manages the values of “failure occurrence date / time”, “model”, “model number”, and “failure code” as in Table D.

  FIG. 13 is a software configuration diagram of the analyst terminal 115 according to the first embodiment. The analyst terminal 115 includes an external data acquisition unit 1301, a transmission instruction request unit 1302, a failure prediction model creation tool 1303, and a failure prediction model implementation tool 1304. The external data acquisition unit 1301 has a function of acquiring sensor data or a failure history from the data collection device 114. The transmission instruction request unit 1302 has a function of requesting the data analyzer 113 to send a sensor data transmission instruction in order to acquire sensor data necessary for creating a failure prediction model.

  The failure prediction model creation tool 1303 is a tool that allows an analyst to create a failure prediction model using sensor data or failure history acquired using the external data acquisition unit 1301. The creation method may be creation by machine learning, expression by a simple mathematical expression, or any other method. The failure prediction model mounting tool 1304 is a tool for mounting the failure prediction model created by the failure prediction model creation tool 1303 in the data analysis apparatus 113. For example, the analyst can create a failure prediction model with the failure prediction model creation tool 1303, and then the analyst can mount the created model on the data analysis apparatus 113 using the failure prediction model implementation tool 1304. The implementation described here is to describe a program, for example.

  As described above, according to the first embodiment, it is possible to reduce the load on the data receiving unit when acquiring data from a large number of image forming apparatuses existing all over the world.

<Example 2>
The failure prediction model applied by the data analysis apparatus 113 described in the first embodiment is not always highly accurate. When a certain failure prediction model is applied to sensor information obtained from a large number of image forming devices all over the world, the failure prediction model easily indicates anomaly against the model implementer's prediction There is. Then, a high-frequency transmission instruction is notified to a large number of image forming apparatuses, and there is a problem that the reception band and the storage area of the data storage apparatus 112 are compressed.

  In order to solve the above problem, a second embodiment will be described with respect to a system that provides an upper limit value of the total number (number) of image forming apparatuses that can simultaneously perform high-frequency transmission. The image forming apparatus exceeding a predetermined upper limit value is stored as a waiting state, and the above-described problem is solved by making a transition from the waiting state to the “high frequency transmission state” at a certain timing. The configuration is the same as that of the first embodiment except for items described below.

  The data managed by the high-frequency transmission history DB 406 in FIG. 4 has the following table F, which is an extension of the table C described in the first embodiment.

  The table C is expanded in that a “waiting” state is added to the “state” column and a “waiting start” column is newly added. The rest of the configuration is the same as that in the table C. The “status” column in Table F represents the status of the image forming apparatus corresponding to the model number, and takes “high frequency transmission completed”, “high frequency transmission in progress”, and “waiting” status. The “waiting start” column represents the start date and time when an image forming apparatus that has been in a waiting state when the total number of image forming apparatuses that are frequently transmitted exceeds an upper limit value.

  The sensor data transmission instruction unit 403 extends the following functions. In addition to the functions described in the first embodiment, the sensor data transmission instruction unit 403 has an upper limit value of the total number of image forming apparatuses that can simultaneously perform high-frequency transmission. When a transmission instruction is notified from the failure prediction unit 401, if the total number of image forming apparatuses that are frequently transmitted and registered in the high frequency transmission history DB 406 exceeds the upper limit value according to the transmission instruction, the image is displayed. The forming apparatus is put in a waiting state and registered in the high-frequency transmission history DB 406.

  The high-frequency transmission history management unit 407 is expanded as follows. In addition to the functions described in the first embodiment, the high-frequency transmission history management unit 407 has an upper limit value of the total number of image forming apparatuses that can simultaneously perform high-frequency transmission. A flow in which the high-frequency transmission history management unit 407 changes the state of the image forming apparatus registered in the high-frequency transmission history DB 406 will be described with reference to the flowchart of FIG.

  FIG. 9 is a flowchart of a processing flow in which the high-frequency transmission history management unit 407 transitions the state of the image forming apparatus registered in the high-frequency transmission history DB 406. The high-frequency transmission history management unit 407 always repeatedly executes Start to End of the flow shown in FIG. 9 in a short cycle. Referring to the correspondence with the sequence diagram of FIG. 11, the flowchart of FIG. 9 is repeatedly executed in a short cycle at an arbitrary timing in the low accuracy failure prediction phase and the high accuracy failure prediction phase.

  First, in step S <b> 901, the high-frequency transmission history management unit 407 determines whether there is an image forming apparatus whose current date and time exceeds the high-frequency stop schedule in Table F managed by the high-frequency transmission history DB 406. . As a result of the determination, if the corresponding image forming apparatus exists, the process proceeds to S902, and if not, the flow ends. In step S <b> 902, the state of the image forming apparatus whose current date and time exceeds the scheduled date and time for high-frequency stop is shifted to high-frequency transmission. The item “status” in Table F is updated, and a high-frequency transmission stop instruction to the image forming apparatus is notified to the sensor data transmission instruction unit 403. In step S903, it is determined whether there is a waiting image forming apparatus. If the image forming apparatus exists, the process proceeds to step S904. If the image forming apparatus does not exist, the flow ends. In step S <b> 904, the state of the image forming apparatus having the oldest “waiting start date / time” among the waiting image forming apparatuses is changed to high-frequency transmission. The item “state” in Table F is updated, and a high-frequency transmission instruction to the image forming apparatus is notified to the sensor data transmission instruction unit 403.

  As described above, according to the second embodiment, it is possible to prevent high-frequency transmission from a number of image forming apparatuses equal to or more than the set upper limit value, and to reduce the reception band and storage area of the data storage apparatus 112.

<Example 3>
As a background, the relationship between the failure prediction system and regular dispatch for maintenance and inspection work by service personnel will be described. If it is determined as a sensor error as a result of the failure prediction in the failure prediction unit 401 in FIG. 4, the sensor error is notified from the data analysis device 113 to the device management device 111, and the serviceman instruction unit 504 in FIG. Notify sensor error. However, when the sensor error is notified to the service person, the sensor is not necessarily broken down. Therefore, the service person who is notified of the sensor error does not always immediately go to the installation location of the corresponding image forming apparatus. . A service person may be dispatched to the installation location of each image forming device regularly. At that time, based on the sensor error notification information, replace or clean the parts that are likely to break down in the near future. Do. For this reason, it is desirable that the failure prediction for the destination image forming apparatus is completed before the service person periodically dispatches.

  In the second embodiment, the high-frequency transmission history management unit 407 changes the image forming apparatus in the high-frequency transmission state registered in Table F to the high-frequency transmission completed state at a high frequency. Performs a transition to the sending state. At this time, since the selection of which waiting image forming apparatus is to be changed to the high-frequency transmission state is made only with reference to the “waiting start date and time”, it is considered from the viewpoint of whether it is in time for the regular dispatch of the service person. The above selection may not always be optimal.

  In order to solve the above problem, in the third embodiment, when the high-frequency transmission history management unit 407 transitions the image forming apparatus from the waiting state to the high-frequency transmission, in addition to the “waiting start date and time”, A system that uses “schedule” (maintenance schedule) as a reference for selection will be described. Since the difference from the second embodiment is only in the flowchart of FIG. 9, FIG. 10 in which the flowchart of FIG. 9 is expanded will be described.

  FIG. 10 is a flowchart of a flow in which the high-frequency transmission history management unit 407 transitions the state of the image forming apparatus registered in the high-frequency transmission history DB 406 in consideration of the regular dispatch schedule of the service person. The high-frequency transmission history management unit 407 always repeatedly executes Start to End of the flow shown in FIG. 10 in a short cycle. Referring to the correspondence with the sequence diagram of FIG. 11, the flowchart of FIG. 10 is repeatedly executed in a short cycle at an arbitrary timing in the low accuracy failure prediction phase and the high accuracy failure prediction phase, as in FIG.

  First, in step S <b> 1001, the high-frequency transmission history management unit 407 determines whether there is an image forming apparatus whose current date and time exceeds the high-frequency stop schedule in Table F managed by the high-frequency transmission history DB 406. . As a result of the determination, if the corresponding image forming apparatus exists, the process proceeds to S1002, and if not, the flow ends. In step S <b> 1002, the state of the image forming apparatus whose current date and time exceeds the scheduled date and time of frequent stoppage is changed to high-frequency transmission. The item “state” in Table F is updated, and a high-frequency transmission stop instruction to the image forming apparatus is notified to the sensor data transmission instruction unit 403. In step S1003, it is determined whether there is a waiting image forming apparatus. If the image forming apparatus is in a waiting state, the process proceeds to step S1004. If not, the flow ends.

  In step S <b> 1004, an image forming apparatus in which high-frequency transmission is completed is extracted from the waiting image forming apparatuses until the next service person is dispatched. In step S1005, the state of the image forming apparatus having the oldest “waiting start date” among the image forming apparatuses extracted in step S1004 is changed to high-frequency transmission. Further, the item “state” in Table F is updated, and a high-frequency transmission instruction to the image forming apparatus is notified to the sensor data transmission instruction unit 403.

  As described above, according to the third embodiment, there is an increased possibility that the failure prediction for the image forming apparatus as the destination is completed before the service person is regularly dispatched, and whether or not the parts are replaced or cleaned when the service person is regularly dispatched. The possibility of being able to make a decision increases.

  In the third embodiment, it has been described that the selection of which waiting image forming apparatus is to be shifted to the high-frequency transmission state is performed in consideration of “scheduled dispatch schedule of serviceman” and “waiting start date and time”. . What information should be taken into consideration can be variously considered from other viewpoints. For example, from the viewpoint of the model creator, if there is a high-frequency model with few applications, feedback on the application of the high-frequency model cannot be obtained. In order to solve this problem, a method of making a transition from the waiting state to the high-frequency transmitting state so that the number of applications of the application model is uniform can be considered.

<Example 4>
The image forming apparatuses 121 to 123 transmit sensor data to the data storage apparatus 112 in response to a transmission instruction from the device management apparatus 111. In the first to third embodiments, since the transmission instruction from the device management apparatus 111 only specifies the transmission frequency and the transmission period, the image forming apparatuses 121 to 123 are other than the data necessary for the failure prediction model applied by the data analysis apparatus 113. There is also the possibility of sending sensor data. In that case, there is a problem of pressing down the bandwidth of the data storage device and the storage capacity.

  In order to solve the above-described problem, in a fourth embodiment, a system in which the type of sensor to be transmitted can be specified regarding the transmission instruction notified by the device management apparatus 111 will be described. Here, the transmission instruction includes a transmission instruction from the device management apparatus 111 to the image forming apparatuses 121 to 123 and a transmission instruction from the device management apparatus 111 to the data analysis apparatus 113. Hereinafter, a configuration different from the first to third embodiments will be described.

  The sensor data transmission instruction unit 403 in FIG. 4 notifies the device management apparatus 111 to transmit the sensor data of the specified sensor type in addition to the specified frequency and period to the specified image forming apparatus (for example, 121). Has function. Depending on how the argument is set, an instruction to stop transmission is also possible. The transmission instruction reception unit 501 in FIG. 5 interprets the contents of the frequency, period, and sensor type specified from the sensor data transmission instruction unit 403, and issues a device transmission instruction to instruct the specified image forming apparatus to perform high-frequency transmission. Part 503 can be requested. In response to an instruction from the transmission instruction reception unit 501 or the sensor error notification reception unit 502, the device transmission instruction unit 503 can directly notify the designated instruction to the designated image forming apparatus (for example, 121). The instruction content to the image forming apparatus includes a sensor data transmission frequency, a sensor data transmission period, a transmission sensor type, and a sensor data transmission stop instruction.

  The transmission instruction reception unit 701 in FIG. 7 has a function of receiving a sensor data transmission instruction from the device management apparatus 111. The instruction content from the device management apparatus 111 includes a sensor data transmission frequency, a sensor data transmission period, a transmission sensor type, and a sensor data transmission stop instruction, and issues an instruction to the sensor data transmission unit 702 according to the instruction content. The sensor data transmission unit 702 has a function of directly transmitting sensor data to the data storage device 112. In response to an instruction from the transmission instruction receiving unit 701, data held in the in-device sensor data management DB 703 can be directly transmitted to the data storage device 112. In accordance with the sensor type notified by the instruction from the transmission instruction receiving unit 701, only the data of the specified sensor type is extracted from the data held in the in-device sensor data management DB 703 and transmitted to the data storage device 112.

  In the flowchart of FIG. 8, only S807 has an extended portion. In S807, a high-frequency transmission instruction is issued. The failure prediction unit 401 in FIG. 4 requests the sensor data transmission instruction unit 403 for the high frequency transmission instruction, and the sensor data transmission instruction unit 403 directly requests the device management apparatus 111 for the high frequency transmission instruction. At this time, the failure prediction unit 401 notifies the sensor data transmission instruction unit 403 of a transmission instruction. The frequency specified in the transmission instruction here uses the value of “input frequency” in Table B corresponding to the application model, and the period specified in the transmission instruction is the value of “input period” in Table B corresponding to the application model. Is used. Further, as the sensor type designated in the transmission instruction, the value of “input sensor” in Table B corresponding to the applied model is used.

  As described above, by performing the above-described expansion, it is possible to issue a transmission instruction that also designates the sensor type, and to reduce the line bandwidth of the data storage device and the compression of the storage area.

<Other examples>
The present invention supplies a program that realizes one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in a computer of the system or apparatus read and execute the program This process can be realized. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

Claims (6)

An analysis system including one or more information processing apparatuses and analyzing data collected from a plurality of network devices,
Determining means for determining a network device to collect data necessary for the second analysis out of the plurality of network devices based on a result of the first analysis;
Control means for controlling the data required for the second analysis to be collected from the determined network device at a frequency higher than the frequency of data collection in the first analysis;
An analysis system comprising: execution means for executing the second analysis using data collected from the determined network device.   In accordance with the result of the first analysis, there is further provided a notification means for notifying a change to the model for the first analysis in accordance with the frequency with which the determined number of network devices reaches a predetermined upper limit. The analysis system according to claim 1, wherein:   The said determination means determines the network device made into the object which collects data required for the said 2nd analysis according to the maintenance schedule of the said network device, The Claim 1 or 2 characterized by the above-mentioned. system.   The analysis system according to any one of claims 1 to 3, wherein the network device includes an image forming apparatus, a digital medical device, a network camera, a robot, an in-vehicle terminal, and an air conditioner.   The analysis system according to any one of claims 1 to 4, wherein the data includes at least one of sensor data, counter information, and log information. An analysis method in an analysis system that includes one or more information processing apparatuses and analyzes data collected from a plurality of network devices,
Determining a network device to collect data necessary for a second analysis from the plurality of network devices based on a result of the first analysis;
Controlling to collect data necessary for the second analysis from the determined network device at a frequency higher than the frequency of data collection in the first analysis;
Performing the second analysis using data collected from the determined network device.

Images