WO2021061116A1

WO2021061116A1 - 3d printer device fleet monitoring

Info

Publication number: WO2021061116A1
Application number: PCT/US2019/052893
Authority: WO
Inventors: Fernando FRIEDRICH; Nailson BOAZ COSTA LEITE; Andre RABELO; Syed Fahad Allam Shah
Original assignee: Hewlett-Packard Development Company, L.P.
Priority date: 2019-09-25
Filing date: 2019-09-25
Publication date: 2021-04-01

Abstract

An example device includes a communications interface and a processor coupled to the communications interface. The processor is to collect data from a fleet of three-dimensional printer devices via the communications interface, set a condition for a performance characteristic of a three-dimensional printer device of the fleet based on a statistical analysis of the collected data, and when the condition is met, communicate an alert indicating the condition to a remote device.

Description

3D PRINTER DEVICE FLEET MONITORING

BACKGROUND

[0001] Three-dimensional (3D) printing and similar types of manufacturing techniques may be used to create diverse objects, such as prototype objects and production objects.

[0002] A three-dimensional printing system may fuse or chemically alter a material, such as powder, to form a printed article. In powder-bed material fusion printing systems, layers of powder are progressively introduced and select portions of each layer are fused with the previous layer. Material fusion may be performed using an energy source, a light source, laser, electron beam, a chemical fusing agent, binding agent, curing agent, an energy absorbing fusing agent, or combination of such that may be jetted or sprayed (e.g., via a thermal or piezo inkjet-type printhead), or similar. Fused layers thereby form a printed article and unfused material may be recovered and recycled.

BRIEF DESCRIPTION OF THE DRAWINGS

[0003] FIG. 1 is a diagram of a 3D printer fleet monitoring device.

[0004] FIG. 2 is a flowchart of a process of monitoring a fleet of 3D printer devices.

[0005] FIG. 3 is a flowchart of a process of monitoring a fleet of 3D printer devices with feedback that modifies monitoring conditions.

[0006] FIG. 4 is a functional block diagram of example architecture of a 3D printer device fleet monitor. [0007] FIG. 5 is a functional block diagram of example architecture of a 3D printer device fleet monitor with a feature table and report generation.

[0008] FIG. 6 shows code of an example alert definition for a 3D printer fleet monitoring device.

DETAILED DESCRIPTION

[0009] A fleet of manufacturing systems, such as thermal-fusion 3D printing systems, chemical binder systems, and/or the like, may be distributed among a large geographical area and may be operated by a variety of different and independent entities under a wide range of parameters and conditions. A given location may include a variety of different 3D printer devices that make up a particular 3D printing system. The range of 3D printer devices available and the possible different configurations of such devices into systems of different capabilities may reduce the effectiveness and increase the complexity of discrete individual system-based monitoring. Further, when systems are similar, redundancy may occur. In other words, different entities monitoring their own systems without regard to the fleet as a whole may lead to reduced efficiency, increased complexity, and unneeded redundancy. As such, a fleet-based monitoring approach may be used.

[0010] The performance of a fleet of 3D printer devices may be monitored and alerts indicative of actual or predicted malfunction may be issued to printer users or a support center. Data about printer performance may be collected and used to trigger alerts that may be communicated to a user’s device or to the support center. An alert may be communicated via an application and related application programming interface (API) or via messaging infrastructure, such as email. Statistical analysis and machine learning may be used to establish an alert, in that data from multiple 3D printer devices operating at different locations under different conditions may be referenced when establishing an alert.

[0011] Feedback from a 3D printer user’s device or a fleet support center may be used to inform the establishment of alerts. Reports of alerts and operations of 3D printing systems may be generated and provided to a fleet support center to inform proactive customer relationship actions and technical support.

[0012] Alerts may be reactive, in that an alert may indicate that an error or other undesirable event/condition has occurred with a specific 3D printer device. Alerts may be proactive and detect a situation that may benefit from support center staff contacting the user of a specific 3D printer device.

[0013] FIG. 1 shows an example 3D printer fleet monitoring device 100. The device 100 may include a server or similar computing device that is connected to a computer network 102. The computer network 102 may include a wide-area network, such as the internet.

[0014] The 3D printer fleet monitoring device 100 monitors the performance of a fleet 104 of 3D printer devices. The fleet 104 of 3D printer devices may include a plurality of 3D printer devices located at different geographic locations. For example, different locations may be operated by different entities (e.g., companies, organizations, schools, etc.) and each location may have a variety of 3D printer devices. A location may have a local-area computer network that may be considered part of the computer network 102.

[0015] Various 3D printer devices may provide different functionality to realize 3D printing at a location. Examples of 3D printer devices include a printer apparatus 110, a trolley 112, and a material processing station 114. A trolley may include a build platform and may be moved between a material processing station 114 and a printer apparatus 110. A material processing station 114 may load a trolley with build or printing material, clean a build platform, recover unused material at the end of a build, and serve other functions. Examples of build or printing material include particulates, powders, liquid/chemical agents, short fiber build materials, binding agents, curing agents, and similar. In some examples a powder may be formed from, or may include, short fibers that may, for example, be cut into short lengths from long strands or threads of material. Build or printing material may include plastic, ceramic, metal powder, and powder-like material, for example. A material processing station 114 may include a processor, memory, and communications interface to carry out its functionality and to communicate with the computer network 102. A printer apparatus 110 may include a mechanism that solidifies printing material, such as by thermal fusing, chemical reaction, and/or by exploitation of other physical phenomena. A printer apparatus 110 may include a processor, memory, and communications interface to carry out its functionality and to communicate with the computer network 102. A printer apparatus 110 may include a printhead, such as a thermal printhead, a piezoelectric printhead, and/or similar device to dispense material or apply heat or other input to material. A given location may have any number and configuration of printer apparatuses 110, trolleys 112, and material processing stations 114. For example, two printer apparatuses 110, five trolleys 112, and one material processing station 114 may be used to implement a specific 3D printing system.

[0016] The 3D printer fleet monitoring device 100 includes a communications interface 120 and a processor 122 coupled to the communications interface 120.

[0017] The communications interface 120 is connected to the computer network 102 and may be connected to a local-area computer network, which supports the 3D printer fleet monitoring device 100 and which may be considered part of the computer network 102.

[0018] The processor 122 may include a central processing unit (CPU), a microcontroller, a microprocessor, a processing core, a field-programmable gate array (FPGA), or a similar device capable of executing instructions. The processor 122 may cooperate with a non-transitory machine-readable medium that may be an electronic, magnetic, optical, or other physical storage device that encodes executable instructions. The machine-readable medium may include, for example, random access memory (RAM), read-only memory (ROM), electrically-erasable programmable read-only memory (EEPROM), flash memory, a storage drive, an optical disc, or similar. The machine-readable medium may store instructions that are executable by the processor 122. [0019] The 3D printer fleet monitoring device 100 further includes 3D printer fleet monitoring instructions 124 that are executable by the processor 122.

[0020] The instructions 124 collect data from the fleet 104 of 3D printer devices via the communications interface 120. Collected data may include part quality indicating data, productivity data, and/or health data. Examples of part quality indicating data include time of last fusion lamp calibration, fusion lamp operational parameters, melt temperature, melt temperature deviation, and similar data that may proxy for build quality. Examples of productivity data include usage (true/false) in past time window (e.g., seven days), usage in deviation from historic/expected, part count in past time window, part count deviation from historic/expected, and similar data. Examples of health data include a high rate of restarts, unexpected shutdowns, a time of last firmware update, and similar.

[0021] The instructions 124 further set a condition for a performance characteristic of a 3D printer device 110, 112, 114 of the fleet 104 based on a statistical analysis of the collected data. Examples of performance characteristics include part quality indicating data, productivity data, and health data, as discussed above, as well as combination of such. For example, a performance characteristic may be “usage” and may be considered normal, low, or high depending on deviation from historic/expected usage. In another example, a performance characteristic may indicate build quality and may include melt temperature and fusion lamp voltage.

[0022] The statistical analysis may include computing an average (mean, median, or mode) or a time-based average (e.g., a mean over fixed duration in the past) and computing a standard deviation or other indicator of variance, then comparing a datapoint to the average and standard deviation. The condition may be a number of standard deviations, such as two, from the average. If a datapoint is outside two standard deviations from the average, the condition is met and the datapoint may be considered an outlier. In other examples, an interquartile range methodology, a k-nearest neighbors methodology, local outlier factor methodology, or other statistical methodology may be used.

[0023] The statistical analysis may include machine learning, as implemented by an artificial neural network, a decision tree, Bayesian network, or similar model. A machine-learning model may be trained in conjunction with operations that use an algorithmic statistical analysis, such as described above. For example, machine-learning model may be trained while production computations are made using standard deviation analysis. Then, once training is completed, the machine-learning model may be used for production operations and the standard deviation analysis may be ceased.

[0024] The condition for the performance characteristic may be set to detect an actual malfunction of a 3D printer device 110, 112, 114. For example, a melt temperature over 90 degrees may indicate a malfunction of a printer apparatus 110. The condition may therefore indicate a performance characteristic that demands a reaction from the operator of the 3D printer device 110, 112, 114 or fleet support center.

[0025] The condition for the performance characteristic may be set to predict a future malfunction. For example, a melt temperature that increases over time at a rate that is unacceptable may be used to predict a future over-temperature state of a printer apparatus 110. The condition may therefore indicate a performance characteristic that is predictive in nature.

[0026] The instructions 124 further determine when the condition is met. When the condition is met, the instructions 124 communicate an alert indicating the condition to a remote device 130, 132. An example remote device is a remote device 130 that belongs to a user of 3D printer device 110, 112, 114. Such a remote device 130 may be part of the 3D printer device 110, 112, 114 or may be an independent device, such as a desktop/notebook computer, tablet computer, or smartphone. Another example remote device is a remote device 132 that is located at a support center associated with the fleet 104. [0027] As such, collected data for a wide variety of 3D printer devices may be accumulated and statistical analysis across the 3D printer devices, as operated at different locations by different entities under potentially different conditions and demands, may be performed to establish performance characteristic alerts for individual 3D printer devices. Collective data for fleet may be used to inform the operation of individual devices, such that a knowledge base much larger than normally generated by an individual device may be applied to the individual device.

[0028] Further, performance characteristic alerts may be set as predictive, so that support center staff may initiate contact with a user of a 3D printer device to address a potential malfunction or other performance degradation before it is realized. For example, support center staff may inform a user that a 3D printer device appears to be operating at too high a temperature and that this may eventually require an on-site technical intervention to repair the 3D printer device. As such, malfunctions and other performance concerns may be preempted.

[0029] FIG. 2 shows an example method 200 of monitoring a fleet of 3D printer devices. The method 200 may be performed with the example 3D printer fleet monitoring device 100 of FIG. 1 or with another device. The method 200 begins at block 202.

[0030] At block 204, data is collected via a computer network from a fleet of disparate 3D printer devices installed at different geographic locations. The 3D printer devices may be operated by different entities. At a given location, 3D printer devices may operate at different stages of a 3D printing process, such as a material loading stage, a printing stage, a material recovery stage, and a cleaning stage. For example, a fleet of 3D printer devices may include printer apparatuses that carry out print operations and material processing stations that load, recover, and clear print material. The fleet of 3D printer devices may further include moveable build platforms, such as provided on trolleys, which may be moved between printer apparatuses and material processing stations. [0031] Data may be continuously collected from the fleet of 3D printer devices over a set time period, such as a day.

[0032] At block 206, a statistical analysis is performed on the collected data. The analysis may also reference data collected prior to the set time period. The statistical analysis may apply a deviation methodology or any of the other statistical methodologies discussed elsewhere herein. The statistical analysis may include application of a machine learning model, as discussed elsewhere herein.

[0033] At block 208, based on the statistical analysis, a condition is set for a performance characteristic of a 3D printer device. The condition may be dependent on the statistical analysis, such as a deviation expressed in terms of a standard deviation, a deviation expressed as a proportion of an average or other reference, a deviation with reference to a time-based average, or similar. The condition may include deviation from an average or similar condition informed by the statistical analysis, as discussed elsewhere herein. The condition may be set for individual 3D printer devices.

[0034] At block 210, it is determined whether the condition has been met.

[0035] If the condition is met, then an alert is communicated via the computer network, at block 212. The alert may be sent to a user of the 3D printer device that triggered the condition or a support center that monitors the entire fleet.

[0036] The method 200 may be continuously executed, so that data is periodically collected and analyzed and so that conditions are continuously tested.

[0037] FIG. 3 shows an example method 300 of monitoring a fleet of 3D printer devices. The method 300 may be performed with the example 3D printer fleet monitoring device 100 of FIG. 1 or with another device. The description for method 200 may be referenced for details not repeated here. The method 300 begins at block 302. [0038] After data is collected from a fleet of 3D printer devices and statistical analysis is performed, at blocks 204, 206, a condition is set, at block 304, based on the statistical analysis.

[0039] The condition may be modified, at block 304, based on feedback received, at block 306, from a remote device, such as a printer user’s device or a support center device, that received an alert triggered by the condition.

[0040] Feedback may include command to ignore an alert, ignore an alert for a certain time, delete/disable an alert, ignore all alerts of the same type, provide a quantified modification to an alert (e.g., melt temperature of 82 degrees is acceptable, update condition to be 85 degrees), or similar. Such feedback may be stored in association with collected fleet data as data that overrides the results of the statistical analysis.

[0041] FIG. 4 shows an example programmatic architecture 400 for a 3D printer device fleet monitor. The programmatic architecture 400 may be used with the device 100 of FIG. 1 and the methods 200, 300 of FIGs. 2 and 3. The programmatic architecture 400 may be implemented by instructions, such as in instructions 124 (FIG. 1), stored in a non-transitory machine readable medium, such as memory 402, and executable by a processor, such as processor 122 (FIG. 1).

[0042] The programmatic architecture 400 may include printer device connection instructions 404, statistical analysis instructions 406, alert generation instructions 408, feedback processing instructions 410, and remote device connection instructions 412.

[0043] The printer device connection instructions 404 manage connections to disparate 3D printer devices at various different geographic locations as well as processes data received via such connections. The printer device connection instructions 404 may provide a server process that accepts remote connections from the 3D printer devices via a computer network, manages such remote connections, receives data provided by the 3D printer devices, and stores such data in association with identify information of the 3D printer devices. The printer device connection instructions 404 may implement an API that is computer- network accessible by instructions executing at the 3D printer devices.

[0044] The statistical analysis instructions 406 implement the selected statistical methodology and/or machine learning model to analyze the data collected by the printer device connection instructions 404.

[0045] The alert generation instructions 408 generate and manage alerts for various 3D printer devices based on the statistical analysis and data received by the printer device connection instructions 404. That is, the statistical analysis instructions 406 may provide a condition for an alert and the device connection instructions 404 may provide data to compare with the condition to trigger or not trigger the alert. The alert generation instructions 408 may further reference feedback data provided by the feedback processing instructions 410 to modify alerts.

[0046] The remote device connection instructions 412 manage connections to various remote devices, such as 3D printer user devices and support center devices. The remote device connection instructions 412 may provide a server process that generates and sends messages containing alerts and/or host an API to which a remote device may connect to obtain alerts via a computer network. The remote device connection instructions 412 may further receive feedback from remote devices.

[0047] The feedback processing instructions 410 process feedback concerning alerts as received from remote devices, such as 3D printer user devices and support center devices. The feedback processing instructions 410 provide relevant feedback data to the alert generation instructions 408 to modify alerts. Alert modification may include ignoring an alert, ignoring an alert for a certain time (e.g., “snoozing” an alert), deleting/disabling an alert, ignoring all alerts of the same type, providing a quantified modification to an alert (e.g., melt temperature of 82 degrees is acceptable, update condition to be 85 degrees), or similar. [0048] The instructions 404-412 may be provided to the same computing device, such as a physical server, or may be distributed across multiple cooperating computing devices, such as a bank of servers or a cloud-based platform.

[0049] FIG. 5 shows an example programmatic architecture 500 for a 3D printer device fleet monitor. The description for the programmatic architecture 400 may be referenced for details not repeated here.

[0050] A feature table 502 may be provided to process data received by the printer device connections 404. The feature table 502 may map collected data to features that may form the basis of statistical analysis and alert generation. Mapping may include reduction of data (e.g., selecting a subset of values from a larger set, combining values into fewer values, etc.), normalization of data from disparate 3D printer devices to a common scale or format (e.g., converting temperature measurements to a common degree scale), computation of time- based values (e.g., average over last 7 days), computation of combined values (e.g., averaging temperature measurements from multiple sensors), generalization of information (e.g., indicating various different computer network errors as “connectivity error”), and similar. That is, the feature table 502 may convert data received from the 3D printer devices, where such data may be different for different devices or unsuitable for creating alerts.

[0051] The data in the feature table 502 may be used by the statistical analysis instructions 406 to generate further statistical data, such as averages, time-based averages, deviations, and similar information, for the features stored in the feature table 502. Such statistical data may also be stored in the feature table 502.

[0052] The data in the feature table 502 may be used by the alert generation instructions 408 to generate and trigger alerts.

[0053] Any number of feature tables 502 may be used. A given feature table 502 may be associated with multiple different alerts. [0054] Configuration instructions 504 may be provided to allow configuration of statistical analysis, alert generation, and features by a user of a 3D printer device or by support center staff.

[0055] Report generation instructions 506 may be provided to generate performance reports for support center staff. Reports may show historic data from the feature table 502 as well as alert history. Reports may also show information related to feedback processing, such as the type of alerts being ignored or modified. Support center staff may reference generated reports when configuring the statistical analysis, alert generation, and features.

[0056] Further, support center staff may receive messages, such as email messages, from an email server 508 implemented by the remote device connection instructions 412. Such messages may include alerts and reports. Independently of such, users of 3D printer devices may receive alerts via application that connects to an API 510 implemented by the remote device connection instructions 412.

[0057] FIG. 6 shows code for example alert definition 600. The alert definition may be created with alert generation instructions 408 (FIGs. 4 and 5). In operation, the alert definition 600 may be applied to a feature table 602 (see also 502 in FIGs. 4 and 5) that is identified in the alert definition 600. An alert condition 604 may be expressed as a function, such as a filter, applied to the feature table 602. A query language may be used to define the alert condition 604. In this example, the alert condition 604 returns the identities of any 3D printer device identified in the feature table 602 that satisfies the query. The alert definition 600 may further indicate a severity indicator 606, so that the recipient of a resulting alert may understand the severity of the alert. Further, the alert definition 600 may include contextual information, such as a title and code describing the alert meaning and intention, as well as source and sensor information, which may help in understanding a root cause of the alert.

[0058] In view of the above, it should be apparent that monitoring a fleet of 3D printer devices may increase efficiency and reduce complexity as opposed to monitoring individual 3D printing systems. A scale of knowledge collected may be greater, so that a particular 3D printing device may have an alert condition that is informed by many other of such 3D printing devices in various configurations and as part of various different 3D printing systems. As such, it is contemplated that fewer false alarms will occur. Further, redundancy that may otherwise exist due to individual operators implementing discrete monitoring systems for their own 3D printing systems may be reduced.

[0059] It should be recognized that features and aspects of the various examples provided above can be combined into further examples that also fall within the scope of the present disclosure. In addition, the figures are not to scale and may have size and shape exaggerated for illustrative purposes.

Claims

1. A device comprising: a communications interface; and a processor coupled to the communications interface, the processor to: collect data from a fleet of three-dimensional printer devices via the communications interface; set a condition for a performance characteristic of a three- dimensional printer device of the fleet based on a statistical analysis of the collected data; and when the condition is met, communicate an alert indicating the condition to a remote device.

2. The device of claim 1 , wherein the processor is further to: modify the condition based on feedback received from the remote device via the communications interface.

3. The device of claim 1 , wherein the processor is further to: set the condition for the performance characteristic to detect an actual malfunction.

4. The device of claim 1 , wherein the processor is further to: set the condition for the performance characteristic to predict a future malfunction.

5. The device of claim 1 , wherein the remote device belongs to a user of three- dimensional printer device.

6. The device of claim 1 , wherein the remote device is located at a support center associated with the fleet of three-dimensional printer devices.

7. A method comprising: collecting data via a computer network from a fleet of disparate three- dimensional printer devices installed at different geographic locations; performing a statistical analysis on the collected data; setting a condition for a performance characteristic of a three- dimensional printer device of the fleet based on the statistical analysis; and communicating an alert via the computer network when the condition is met.

8. The method of claim 7, wherein the fleet of disparate three-dimensional printer devices comprises devices to operate at different stages of a three- dimensional printing process.

9. The method of claim 7, wherein the fleet of disparate three-dimensional printer devices comprises printer apparatuses and material processing stations.

10. The method of claim 7, further comprising modifying the condition based on feedback received from a recipient of the alert.

11. The method of claim 7, wherein performing the statistical analysis on the collected data comprising applying a machine learning model.

12. A non-transitory machine-readable medium comprising instructions that cause a processor to: collect data via a computer network from a fleet of disparate three- dimensional printer devices installed at different geographic locations; perform a statistical analysis on the collected data; set condition for an alert based on the statistical analysis; and trigger the alert when the condition is met.

13. The non-transitory machine-readable medium of claim 12, wherein the instructions are further to modify the condition based on feedback received from a recipient of the alert.

14. The non-transitory machine-readable medium of claim 12, wherein the instructions are further to process the collected data with a feature table.

15. The non-transitory machine-readable medium of claim 14, wherein the instructions are further to set the condition with reference to the feature table.