Work result prediction system, method, device, and service provision system

Abstract

To provide a work result prediction system, a device, and a service provision system that enable work results to be easily predicted.SOLUTION: In an embodiment, a worker produces products by time series movement to a plurality of at-desk positions in a work area photographed by a camera. A storage unit stores appearance number data of a good at-desk time series pattern obtained by presetting the good at-desk time series pattern having good manufacture number data among a plurality of types of at-desk time series patterns in which the worker moves to the at-desk positions in time series, and the manufacture number data of the products corresponding to the appearance number data in association with each other. A pattern appearance detection unit measures the good at-desk time series pattern when the products are actually manufactured and obtains the appearance number data. A prediction unit identifies the manufacture number data stored in the storage unit based on the appearance number data and obtains a predicted manufacture number of the products corresponding to the manufacture number data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a work performance prediction system, a device, and a service provision system. [Background technology]

[0002] There are production, manufacturing, or construction sites that have multiple work sites (multiple seating positions) within a production site (for example, within a single booth or a pre-defined area). In this case, for example, one worker moves to the multiple work sites and performs the necessary work at each site. For example, the following work process can be considered as an example of production work within a single booth.

[0003] a first work process in which, for example, two parts are assembled at a first work site to obtain one composite part; a second work process at a second work site for adjusting, for example, a part position or a property of the composite part; a third work step of mechanically or electrically testing the adjusted composite part at a third work site; At the fourth work site, the tested composite part is loaded onto a conveyor belt and is in a waiting process where it waits to pick up the next two parts to be assembled.

[0004] In this type of work, the number of finished products produced in a day varies greatly depending on the experience and skill of the worker.

[0005] For example, to understand the number of products completed in a day (i.e., work results), factory managers compile data reporting the number of products completed by booth workers. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2020-95651 [Patent Document 2] International Publication No. 2019 / 039126 Summary of the Invention [Problem to be solved by the invention]

[0007] There are multiple booths like the one mentioned above set up in the factory. To predict the number of products completed in a day, the manager tallies the number of completed products at each booth, as recorded in the reports of the workers at each booth. However, with this method, if there is an error in the report, an error will occur in the total number of completed products. The error will also increase as the number of booths increases. Also, it takes time to collect the reports and obtain the total number of completed products (work results) (this method is not very immediate).

[0008] Therefore, in one embodiment, the object is to provide a work result prediction system, device, and service provision system that makes it possible to easily predict work results in an environment where, for example, there are multiple work sites within a set area and work results are obtained by workers moving between the multiple work sites and completing each task at each site. [Means for solving the problem]

[0009] According to one embodiment, a plurality of work positions are set within a predetermined work area, and a work environment is provided in which a worker can move to the plurality of work positions and perform work to obtain a work result, The apparatus includes a storage unit, a pattern appearance detection unit, and a prediction unit, The storage unit stores occurrence count data of a good occupation time series pattern that has good performance data of the work result, the good occupation time series pattern being preset among a plurality of types of occupation time series patterns that are formed by the worker moving to a plurality of the work positions, and the work result data of the work result corresponding to the occurrence count data. and remember it. the pattern appearance detection unit acquires the appearance count data of the good occupancy time series pattern from the occupancy time series patterns detected when the work is actually performed, A work result prediction system is provided in which the prediction unit identifies the occurrence count data stored in the memory unit based on the occurrence count data acquired by the pattern occurrence detection unit, and obtains the result data corresponding to the identified occurrence count data as the work result. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram specifically showing an embodiment of the present invention. [Figure 2] FIG. 2 is a block diagram showing the main part of an embodiment of the present invention. [Figure 3] FIG. 3 is an explanatory diagram showing an example of raw data of a seating time series pattern when the production efficiency of the product is not good in the seating location detection system. [Figure 4] FIG. 4 is an explanatory diagram showing an example of raw data of a seating position time series pattern when the production efficiency of the product is good in the seating position detection system. [Figure 5] FIG. 5 is an explanatory diagram showing an example of a rule for determining seat location data in one embodiment of the present invention. [Figure 6] FIG. 6 is an explanatory diagram showing a modified version of the seating time series pattern shown in FIG. [Figure 7] FIG. 7 is an explanatory diagram showing a modified version of the seating occupancy time series pattern shown in FIG. [Figure 8] FIG. 8 shows an example of a good occupied seat time series pattern indicating the order of occupied seat positions adopted in the system of this embodiment. [Figure 9] Each of (a), (b), (c), and (d) in FIG. 9 is an explanatory diagram showing the correlation between the number of occurrences (horizontal axis) of the time series pattern of seating positions and the number of productions within a given time period. [Figure 10] FIG. 10 corresponds to FIG. 6 and is an explanatory diagram showing an example in which information on production numbers is visualized and displayed. [Figure 11] FIG. 11 corresponds to FIG. 7 and is an explanatory diagram showing an example in which information on production numbers is visualized and displayed. [Figure 12]FIG. 12 is a block diagram showing another embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments will be described with reference to the drawings. An embodiment is shown in Fig. 1. In this embodiment, for ease of explanation, an example will be described in which the work area is a booth and a worker produces a product.

[0012] The reference numerals indicate one booth in schematic form, and show that multiple (for example, four) work sites P1, P2, P3, and P4 are installed within this booth 200. For example, one worker is in charge of one booth 200. To produce one product, this worker moves between the work sites P1, P2, P3, and P4, and produces the product in stages at each work site (work including obtaining parts, assembling, adjusting, testing, etc.).

[0013] For example, a 360-degree camera 300 is placed above booth 200. The entire work sites P1, P2, P3, and P4 within booth 200 are within the field of view of 360-degree camera 300 (hereinafter referred to as camera 300), and can be captured by camera 300. Use of camera 300 offers the convenience of not having to adjust or reposition the 360-degree camera each time the layout of the work sites within booth 200 is changed or the number of work sites is increased or decreased.

[0014] In the above-mentioned booth 200, a 360-degree camera 300 was used for monitoring. However, since the purpose of this camera 300 is to detect when workers are at work sites P1, P2, P3, and P4, many other means are possible to achieve this purpose. For example, one method is for a sensor device attached to a desk or chair at each work site P1, P2, P3, and P4 to communicate with a wireless tag or IC card carried by the worker to detect entry and exit.

[0015] An example of work performed by a worker in booth 200 will be described below.

[0016] At the first work site P1, for example, a first operation is performed in which two parts are assembled to obtain one composite part (assembly process). At the second work site P2, a second work is performed to adjust, for example, the component position or characteristics of the composite component (adjustment step). At the third work site P3, a third work is performed in which the adjusted composite part is mechanically or electrically tested (testing step). At the fourth work site P4, the tested composite part is loaded onto a conveyor belt, and a fourth operation is performed in which the next two parts to be assembled are picked up (waiting process).

[0017] To produce one product, a worker performs each task (part assembly, adjustment, testing, part picking, etc.) at each of the above work sites (which can be called seating positions) P1, P2, P3, and P4 in sequence to produce one product.

[0018] However, during actual production, for example, during the testing (main work) stage, there may be a slight error in the electrical adjustment of the product. In such a case, the worker may return to the designated work site to correct the error and perform electrical readjustment (auxiliary work). Also, during product assembly adjustment (main work), a worker may notice deformation of a part. In this case, the worker may return to the part assembly work site and replace the part (auxiliary work). Also, for some reason, there may be cases where the worker is not present at any of the work sites (in this case, let's call this work site P0).

[0019] As described above, in a work process, there are cases where a worker travels back and forth between multiple work sites in a short period of time.

[0020] Additionally, a single worker in a booth may handle multiple units of the same type of product while performing production tasks. For this reason, it is possible that a worker may be performing assembly work for a second product while adjusting or testing the first product. In other words, parallel processing of tasks may occur. In this case, the worker's main task is adjustment or testing, while also performing assembly work as a secondary task. Furthermore, there are various combinations of work sites for main and secondary tasks.

[0021] The above work is captured by camera 300. The captured image signal from camera 300 is input to stream receiver 401 of interface 400. Interface 400 also has a recording device 402, and the captured image signal received by stream receiver 401 is input to recording device 402 as a video signal, which is recorded and saved. The video signal recorded in recording device 402 is displayed on a display and may be used in the future for inspection, confirmation, training, or learning.

[0022] The video signal output from the stream receiver 401 is input to a capture 501 of a time series pattern appearance detection unit 500 .

[0023] The time series pattern appearance detection unit 500 basically includes a capture 501 and an image processor 502, and further includes a memory 503 for storing presence detection data and staying time data, a processor 504 for converting to presence time series pattern data, a memory 505 for presence time series pattern data, a processor 506 for extracting good presence time series pattern data, a memory 507 for specific pattern data, and a memory 508 for the appearance frequency of good presence time series pattern data (which may also be simply referred to as memory 508).

[0024] The combination pattern data of P1, P2, P3, P4, and P0 may be the pattern data (combination data from P0 to P4) itself indicating the time series pattern. Alternatively, it may be an identification number of the combination pattern from P0 to P4. This identification number is numerical data that identifies the combinations from P0 to P4 by exhaustively searching.

[0025] The capture 501 receives a video signal from the stream receiver 401 and outputs the video signal to the image processor 502 .

[0026] The image processor 502 (A1) detecting a worker in the video signal; (A2) Determine which location in the booth the worker is at (P1, P2, P3, P4, or a different location P0), and then: (A3) The time the worker stayed at the determined seating location (stay time) can also be recorded. Note that in this explanation, it is assumed that one worker works in one booth, but there are also cases where multiple people work together. Also, although five sites were used as an example in the above example, the number of sites is not limited.

[0027] Specifically, the information about the worker's working time in the image processor 502 is detected as follows: (B1) For example, a specific brightness area (brightness area of ​​face, clothes, helmet, etc.) that indicates a worker moving within the booth is detected (the worker is detected on the image). In this case, it is desirable that the clothes, helmet, etc. worn by the worker are of a color that does not provide camouflage within the booth. When a specific brightness area is detected at a seating position (P1, P2, P3, P4, or a different position P0), counting of the time spent at that seating position begins. Furthermore, when multiple workers are working, each worker is identified and detection processing is performed for each worker.

[0028] (B2) Next, when the specific brightness area (worker) moves to another position, the time spent there (stay time) is determined. Then, the location of the person at the new position after the move is determined, and the time spent there is measured.

[0029] (B3) The process of (B2) above is repeatedly executed.

[0030] The method for detecting workers is not limited to detecting specific brightness areas, but may also include detecting specific color areas or specific shape areas. Furthermore, a human sensor may be used to detect specific temperatures, either independently or in combination. Furthermore, worker tracking may be performed, using image processing (circuitry or software) to detect motion vectors.

[0031] The image processor 502 sequentially obtains data on the seating position and the duration of stay, and these data are temporarily stored in the memory 503 .

[0032] The image processor 502 differs depending on the type and content of the work, but here, the determination (detection) of the worker's seat position is performed at a frame rate of, for example, 5 frames per second or more, and it is determined at one of the seat positions P1, P2, P3, P4, and P0 at one-minute intervals whether the worker is located (present). In other words, the seat position is detected continuously, but the seat position is ultimately determined as data at one-minute intervals. This is because a worker may work at one seat position, move to another work position, and return within a short period of time. The frame rate and determination interval may be changeable by the system administrator depending on the type and content of the work.

[0033] That is, the system is designed so that the frame rate for detecting the seat location and / or the cycle for determining (deciding) the seat location can be changed by the administrator.

[0034] The image processor 502 first supplies the detected raw seat position data to the memory 503. The image processor 502 also measures the stay time data of the worker who stayed at each seat position P1, P2, P3, P4, or P0, and supplies this stay time data to the memory 503. The seat position data and stay time data are stored in the memory 503. The memory 503 also stores time data indicating the acquisition order of the seat position data and stay time data as a set.

[0035] Next, the occupied position data indicating the occupied position in the memory 503 and the data on the stay time when the person stayed at that position are read out and processed by a conversion processor (which may be called a time series conversion processor) 504 to convert the data into occupied time series pattern data.

[0036] The time series conversion processor 504 identifies the seating location data while referring to the staying time data in the order in which the raw seating location data was acquired, and converts the identified seating location data into time series data.

[0037] The time series conversion processor 504 at this time uses a method specific to this system, and the basic processing method of this method will be explained later.

[0038] The reason for using this unique method is that workers frequently move to multiple auxiliary positions that are different from their original work position (the position of the main work). Movements to such auxiliary work positions are called auxiliary movements. Auxiliary movements can be for readjustment of products, adjustment of the assembly position of parts, testing, simple equipment overhaul, or cleaning. In other words, main work and auxiliary work can be performed in parallel, resulting in complex movements (position changes).

[0039] For this reason, when identifying the final location of a worker, the time series conversion processor 504 identifies the location where the worker has been present for a long period of time or the location where the worker has been present frequently as the location. In this case, the time spent by the worker at the identified location is identified as the time (ignored time) excluding the time spent by the worker when the worker moved to another location (auxiliary movement). In other words, a determination processing function (which may also be called a smoothing processing function or a rectification processing function) is provided to identify the time spent by the worker when the worker moved to the original work location, assuming that the auxiliary movement did not occur. This is because the time series conversion processor 504 aims to determine the time series pattern of multiple locations where the worker performed actual work (original main work). Note that actual daily time data indicating the order in which the location data and the time spent data were acquired are also stored as a set. The time series conversion processor 504 also includes a processor or program for performing the above processing.

[0040] In determining the above-mentioned occupied position, in the above example, the position with the highest frequency is set as the occupied position, but the last occupied position (the position to which the worker last belonged) may also be set as the occupied position, and this can be set appropriately on-site. This is because when a worker moves from a predetermined position among positions P1, P2, and P3 to another position, the worker often moves a part or composite part from that predetermined position to the next process.

[0041] The seat location data indicating the seat location and the data on the duration of stay at the location are input to a time series conversion processor 504. This conversion processor 504 arranges the seat location data (the seat location identified every minute) in a time series.

[0042] That is, the time series conversion processor 504 acquires the seating position data obtained in chronological order from the memory 503, detects the stay time data at the seating positions in ascending order, and arranges the seating position data corresponding to each stay time data in chronological order. When determining the time order in this way, the time data at the time of stay detection attached to the data indicating P0, P1, P2, P3, and P4 is referenced. Furthermore, at this time, the seating position data and stay time data for seating positions P4 and P0 are cut because they do not contribute to production work.

[0043] Therefore, the time series conversion processor 504 creates, for example, a time series pattern in which three items P1, P2, and P3 are arranged in a time series according to their respective stay times (stay order), or a time series pattern in which six items P1, P2, P3, P1, P2, and P3 are arranged in a time series according to their respective stay times (stay order).

[0044] The time series pattern data created by the time series conversion processor 504 is stored in a memory 505 for seating presence time series pattern data (sometimes simply referred to as a memory).

[0045] Next, the plurality of pieces of occupied time series pattern data (detected data) stored in memory 505 are supplied to a good occupied time series pattern data extraction processor (sometimes referred to as a pattern data extraction processor) 506. The pattern data extraction processor 506 is provided with good occupied time series pattern data (previously identified and prepared data) for reference from a memory 507 (sometimes simply referred to as memory 507) of specific (may be referred to as a good time series) pattern data.

[0046] Here, pattern data extraction processor 506 compares the plurality of occupancy time series pattern data provided from memory 505 with the good occupancy time series pattern data provided from memory 507. Then, it determines how many pieces of pattern data identical to the good occupancy time series pattern data exist in the plurality of occupancy time series pattern data (detected data), for example, in the morning, afternoon, and throughout the day. In other words, it determines the number of occurrences (which may also be referred to as the number of times or frequency) of the good occupancy time series pattern data existing in the detected plurality of occupancy time series pattern data. Then, data indicating this number of occurrences is stored in memory 508.

[0047] The data indicating the number of occurrences is given to a production number prediction data calculator 602 in the production number prediction unit 600. The production number prediction data calculator 602 is connected to a memory 601 having a correlation (linking) table between good occupancy time series pattern data and production number data of products (hereinafter referred to as a production number prediction table).

[0048] The number of occurrences of the above-mentioned good attendance time series pattern is, for example, the number of occurrences after 2 hours or 3 hours have elapsed. This set time of 2 or 3 hours is not fixed, and can be changed as appropriate by the system administrator depending on the product to be produced and the date and time. If the desired production quantity is a product that can be produced in a short time (about 5 hours), the set time can be shortened, and if the product can be produced in a full day, the set time can be longer.

[0049] The memory 601 stores a production volume prediction table acquired in advance (by deep learning). This table stores production volume data according to the number of times a good occupancy time series pattern appears per set time period. The production volume data according to frequency is prepared as, for example, a predicted production volume for the morning, a predicted production volume for the afternoon, and a predicted production volume for the day.

[0050] The production quantity prediction data calculator 602 uses data indicating the number of occurrences (frequency) and the production quantity data of the product from the memory 508 to perform calculations to predict, for example, the production quantity of the product for the morning, afternoon, and daily. Note that although the above description has been given with one booth 200 in mind, there may be multiple booths 200. When there are multiple booths 200, the data acquired at each booth may be processed in a time-sharing manner, or a system may be provided for independently processing the data acquired at each booth. In such a case, the production quantity prediction data calculator 602 is provided with a function for summing up the predicted production quantities of the multiple booths. Alternatively, a tallying unit may be provided for tallying the production quantity data from the multiple production quantity prediction data calculators 602 located at each booth.

[0051] The production quantity forecast data calculator 602 or the above-mentioned tabulator can convert the forecast production quantity into visualized data and provide it to the display device 800. If the calculated forecast production quantity does not reach the target value, a warning signal is provided to the alarm device 700 to notify the system user of the warning. The display device 800 uses a personal computer and is integrated with an operation unit 801 (keyboard).

[0052] The display 800 and operation unit 801 may be arranged in a separate monitoring room or a surveillance room.

[0053] In the above description, the memory 601 is described as storing a production quantity prediction table that has been checked in advance.

[0054] The system may have a function to automatically create a production quantity prediction table. The automatic creation function may be performed during a preliminary run to produce a product on a trial basis. Alternatively, the system supplier may perform a trial run in advance to create a production quantity prediction table.

[0055] The production quantity prediction table creator 900 (shown in the upper right corner of FIG. 1) will be described below.

[0056] The production number prediction table creator 900 includes a memory 902 for storing occupancy time series pattern data, a production number data holder 903, and a detector 904 for occupancy time series pattern data when the production number is large.

[0057] The memory 902 for storing the seating occupancy time series pattern data holds the same data as the memory 505 for storing the seating occupancy time series pattern data. Therefore, when the present system is in the operation mode for creating a production number prediction table, the memory 505 may be substituted for the memory 902.

[0058] The production number data holder 903 sequentially receives production number data from the product counter 1110 during a preset time period. It also stores the time at which each production number data was received. Then, for example, it holds the production number data per preset time period and arranges them in chronological order.

[0059] On the other hand, the occupancy time series pattern detector 904 determines a plurality of pieces of production number data in descending order of production number from the plurality of pieces of production number data held by the production number data holder 903, acquires the production number data, and temporarily stores the data in memory. At the same time, the occupancy time series pattern detector 904 also acquires data (referred to as good production time data) of the time (corresponding to the production number data) when the plurality of pieces of production number data determined by the production number data holder 903 were obtained, and temporarily stores the data in memory.

[0060] The system administrator can set in advance the above-mentioned plurality of production volume data in descending order of the number of production volumes.

[0061] Next, the detector 904 for the occupancy time series pattern refers to the production time data linked to the occupancy time series pattern stored in the memory 902, and identifies the data that matches the good production time data stored in the detector 904. Then, the detector 904 also identifies the occupancy time series pattern (in the memory 902) that corresponds to the identified good production time data, creates a table, and transmits this data to the memory 601 for storage. As a result, the memory 601 stores the production quantity prediction table that has been previously investigated.

[0062] Each block of the above system may be configured as hardware with its own dedicated data processor. A system controller 1111 controls all the blocks via a bus line (overall control of setting the processing order for each block, starting, pausing, data transfer, etc.). Each block may also be configured as a software-based data processing block. The system controller 1111 may be configured to control the operation timing of each block.

[0063] FIG. 2 shows, in large blocks, the overall configuration of the production equipment equipped with the production evaluation system shown in FIG. 1. That is, the production equipment equipped with the production evaluation system is A camera 300 is placed above a booth 200 equipped with multiple work sites. An image signal from the camera 300 is input to a time series pattern appearance detection unit 500 via an interface 400. This time series pattern appearance detection unit 500 includes a memory 507 and a production quantity prediction unit 600.

[0064] The memory 507 stores pattern identification data for identifying a good occupancy time series pattern that can differentiate the production efficiency of the product among a plurality of occupancy time series patterns in which workers move to their occupancy positions in time sequence, in association with the production quantity data of the product corresponding to the good occupancy time series pattern. Furthermore, the time series pattern occurrence detection unit 500 measures the occurrence of the good occupancy time series pattern among the plurality of occupancy time series patterns when the product is actually produced. The production quantity prediction unit 600 obtains the predicted production quantity of the product using the measurement data of the good occupancy time series pattern measured by the pattern occurrence detection unit 500 and the production quantity data.

[0065] 3 and 4 are diagrams for explaining the method for obtaining the above-mentioned seat location data (or work location data) and stay time data. In each diagram, the vertical axis indicates the seat location, and the horizontal axis indicates the time when the seat location was determined (time within the system).

[0066] Fig. 3 shows raw data when production efficiency was poor, and Fig. 4 shows raw data when production efficiency was good. Figs. 3 and 4 show the detection status of occupied positions P0, P1, P2, P3, and P4 for one day. This data is the data stored in memory 503 for occupied position data and staying time data described in Fig. 1.

[0067] Fig. 5 is an explanatory diagram showing the basic rules for detecting seat positions P0, P1, P2, P3, and P4 and specifying seat position data. Fig. 5 is a schematic diagram for ease of understanding. Fig. 5 shows raw seat position data.

[0068] The vertical axis indicates the occupied positions P0, P1, P2, P3, and P4, and the horizontal axis indicates the time when the worker was detected to be at each work site, ranging from 0 to 30 minutes in 1-minute intervals. The occupied position data is identified every minute. For example, if a worker is detected at occupied position P0 and remains there for 1 minute, the worker is determined to have been at occupied position P0, and the occupied position data P01 is identified.

[0069] Furthermore, as shown in the figure, even if a worker travels back and forth between seat position P1 and seat position P2 within a given minute (between 7 and 8 minutes), the seat position P1 with the higher seat position detection frequency is identified, and seat position data P11 is identified as a rule.

[0070] This means that during the one minute period (between seven and eight minutes) in Figure 5, the frequency of presence at seat position P1 was higher than the frequency of presence at seat position P2. This identifies seat position data P11 for one minute. Furthermore, because the worker worked at seat position P1 for two minutes, seat position data P12 is identified. The fact that the worker moved to seat position P2 may indicate that some kind of side work was performed.

[0071] Next, the worker moves to seat position P2 and works for one minute (occurrence of seat position data P21), then returns to seat position P1 and performs assembly work (occurrence of seat position data P13), moves again to seat position P2 and performs adjustment work (occurrence of seat position data P22), and then returns to seat position P1 and performs assembly work for two minutes (occurrence of seat position data P14).

[0072] After 15 minutes have passed, the worker moves to work position P3 and performs test work for one minute (occupancy location data P31 is generated). Immediately thereafter (16 minutes have passed), the worker is detected at work position P4, but because this action lasted less than one minute, no seat location data is generated. In this example, it is predicted that the worker did not hesitate to select the next part and immediately selected it.

[0073] After 16 minutes, the worker is detected at work position P1 and works for two minutes (occurrence of seat position data P15). Then, after 18 minutes, the worker moves to work position P4. Here, it is predicted that the worker may have forgotten a part that he was supposed to pick up, and he works for one minute at work position P4 (occurrence of seat position data P42). At 19 minutes, he returns to work position P1 and performs assembly work (occurrence of seat position data P16). The next seat position data P23 (corresponding to adjustment work), seat position data P32 (corresponding to test work), and seat position data P17 (corresponding to assembly work) indicate that the work was completed smoothly.

[0074] In this system, when generating an occupation time series pattern from the raw occupation location data, the occupation time series pattern conversion processor 504 performs the following special processing. Specifically, the occupation location data for occupation locations P0 and P4 in areas that do not contribute to the production process are deleted. For this reason, the occupation location data P01, P41, and P42 in Figure 5 are deleted. Furthermore, the stay time data at occupation locations P1, P2, P3, and P4 is also compressed to, for example, the smallest unit. Therefore, the concept of time units shown in Figure 5 disappears, and the data is converted into information indicating whether or not work was performed while moving. In other words, this is time-compressed. Alternatively, the occupation location data can be considered information indicating movement changes to work locations that contribute to production, omitting the stay time at occupation locations P1, P2, P3, and P4. In the example shown in Figure 5, this information can be expressed as a position pattern: 121213123121123... Or, it can be said that the seat location data (1 minute = 1 data x 30 minutes) is compressed into 14 data. By performing the above processing, it is possible to significantly compress the amount of data recorded in memory 505.

[0075] Of course, together with the compressed data, the location data of the same location may be extracted. This data is useful for managing the time spent at each work site (stay time).

[0076] Figures 6 and 7 show examples of editing (processing) the seat location data in Figures 3 and 4 using the method described above. This simplifies the seat location data to enable efficient judgment of production efficiency. To achieve this, data that does not significantly affect production efficiency is cut out.

[0077] That is, in Figure 6, when the worker moves to positions P0 and P4, it does not contribute significantly to increasing production efficiency. Therefore, the data on the worker's seating position when the worker moves to positions P0 and P4 has been deleted. Therefore, in this case, the horizontal axis (time) of the plotted data is compressed. Naturally, this also has the effect of reducing the amount of data in the memory that stores this data.

[0078] In Fig. 7, the data on the occupied positions when the worker moved to positions P0 and P4 has been omitted, as in Fig. 6. In this case too, the horizontal axis (time) of the plotted data has been compressed, resulting in a reduction in the amount of data.

[0079] Even when the data conversion process is performed as described above, it can be seen that the worker moves between positions P1, P2, and P3 more frequently in the case of Figure 7 than in Figure 6. In other words, the vertical bars in the graph showing the movement determination results in the figure are denser in the time direction.

[0080] Furthermore, when creating the seat location data of Figures 6 and 7, as mentioned above, this system identifies the seat location of a worker for one minute by identifying the position where the worker was seated for a long period of time or the position where the worker was seated frequently. In the above data processing, if the seat location for one minute is determined, and the same seat location is determined for the next minute as well, the stay time is determined as if the worker was seated in the same position for two minutes.

[0081] Therefore, this system not only simply measures the seat location and the time spent there, but also makes it possible to use the time spent there as reference data for later use. For example, if the time spent at the same location varies depending on the worker, the time spent there can be used as a basis for comparing the worker's proficiency, work aptitude, and work ability, and can also be used to judge the setting status of the equipment at the work location.

[0082] Furthermore, looking at the worker's movements from 7 minutes to 8 minutes in Figure 5, the worker moves from position P1 to position P2 and then back to position P1 within one minute. However, because the presence frequency (i.e., presence time) at position P1 is high, the determination determines the presence position P1 and identifies the presence position data P11. When storing the presence position data P11 as elapsed time data P11b, the elapsed time data P20b at position P2 may be subtracted from the elapsed time data P11b (i.e., time-compressed) and stored. This is because it is effective in reducing the amount of data when processing the distribution data of the presence time series pattern. However, it is preferable to separately record and save the elapsed time data P20b at position P2 rather than completely discarding it. This is because it is useful for later investigation of the factors that led the worker to move to position P2. Figure 8 shows an example of a time series pattern indicating the order of work positions used in this system. When monitoring this time series pattern, an example of a pattern using two sets of positions P1, P2, and P3 is shown. In this embodiment, the purpose can be achieved with one set instead of two sets.

[0083] Positions P1, P2, P3, P1, P2, P3 are used to determine time series combinations through a round-robin approach, and a pattern identification number is assigned to each combination. Four types of pattern identification numbers are shown in the figure: 304, 183, 551, and 195. The combinations of pattern identification numbers and time series patterns with the highest correlation are shown from top to bottom in the figure. In the case of pattern identification number 304, the time series pattern is P2, P1, P3, P1, P3, P2. For pattern identification number 183, the time series pattern is P1, P3, P1, P3, P2, P1. In the case of pattern identification number 551, the time series pattern is P3, P1, P3, P2, P1, P3. For pattern identification number 195, the time series pattern is P1, P3, P2, P1, P3, P1.

[0084] In the embodiment, the production efficiency of the products was excellent when the above four time series patterns were used. Therefore, the four time series patterns were selected as representative patterns in descending order of correlation between the time series patterns and the production efficiency of the products and are shown in the figure.

[0085] In this embodiment, the time series pattern with pattern identification number 195 is designated as the good occupied time series pattern. This designation can be arbitrarily set by, for example, a factory manager, and is not limited to this, and any of pattern identification numbers 304, 183, and 551 may also be used.

[0086] The first approach is to avoid incorporating human preconceptions and assumptions, and to use the time series pattern with the highest correlation among the extracted time series patterns.

[0087] The second approach is that when there are multiple production lines with multiple booths lined up in parallel, a common time series pattern may be adopted from among multiple time series patterns that are good candidates for each production line.

[0088] As a third approach, when there are multiple time series patterns that are good candidates collected over different time periods, a common time series pattern across the different time periods may be used.

[0089] A fourth approach is that when multiple time series patterns are obtained for different workers, a time series pattern that is common to the workers may be adopted.

[0090] Each of (a), (b), (c), and (d) in FIG. 9 shows the correlation between the number of occurrences of a time series pattern (horizontal axis) and the production volume within a predetermined time period (for example, 8 hours). FIG. 9(a) plots the correlation between the number of occurrences of the time series patterns P2, P1, P3, P1, P3, and P2 of pattern identification number 304 (shown in FIG. 8) and the production volume. Figure 9(b) plots the correlation between the number of occurrences of the time series patterns P1, P3, P1, P3, P2, and P1 of pattern identification number 183 (shown in Figure 8) and the number of productions. Figure 9(c) plots the correlation between the number of occurrences of the time series patterns P3, P1, P3, P2, P1, and P3 of pattern identification number 551 (shown in Figure 8) and the production volume. FIG. 9(d) plots the correlation between the number of occurrences of the time series patterns P1, P3, P2, P1, P3, and P1 of pattern identification number 195 (shown in FIG. 8) and the production volume.

[0091] In both cases, the plots show a strong correlation between the number of occurrences of the time series pattern P1, P3, P2, P1, P3, P1 and the production numbers.

[0092] Using the data in Figure 9 above, the production volume (predicted value) can be predicted as follows: Using the data in Figure 9, a simple regression line is calculated using the following formula: First, the slope α and intercept β can be calculated. Slope α = correlation coefficient × (standard deviation of production numbers) / (standard deviation of pattern occurrence counts) Intercept β = average number of productions - (slope α × average number of pattern turns) Production volume (predicted value) = slope α × number of pattern occurrences (measured value) + intercept β The predicted value can be obtained by

[0093] Therefore, if the number of times a pattern occurs (actual measurement value) is known, it is possible to predict the number of products produced. This predicted data is used as report information, and as explained next, it can also be used as visualized information in near real time (for example, every 155 minutes).

[0094] 10 and 11 are basically the same as those shown in FIGS. 6 and 7, and the data is stored in memory 505. FIGS. 10 and 11 show an example in which, using the data in memory 505, marks (colors, symbols, etc.) (T01, T11, T12, T13, T14) indicating that the production of the product is proceeding smoothly are simultaneously displayed on display 800 in accordance with a time series pattern, in accordance with the number of times the set pattern appears. This marks indicates the timing of appearance, marking the timing of appearance based on the number of times the set pattern appears. This allows the user to determine at a glance whether the production situation is good or bad.

[0095] That is, this system can monitor the occurrence of a predetermined time series pattern on the display 800 in the monitoring room, and can estimate the production volume from experience.

[0096] FIG. 12 is a block diagram showing another embodiment. The same reference numerals as in FIG. 1 are used to denote components corresponding to those in the embodiment of FIG. 1. In this embodiment, an image signal from the camera 300 is temporarily stored in the image buffer memory 411 of the interface 400. Multiple pieces of preset information can be input to the preset information storage unit 420 from an operation unit (e.g., the operation unit 801 of the display 800 shown in FIG. 1). The preset information is, for example, information for specifying the aforementioned seat positions P1, P2, and P3. Using the preset information, the cropped image creation unit 412 can retrieve image data from the image buffer memory 411 and crop an image of only the required area. The cropped image data is then temporarily stored in the cropped image buffer memory 413. This reduces the amount of data required to determine whether or not a worker is seated at least at the seat positions P1, P2, and P3.

[0097] The image data in the cropped image buffer memory 413 is input to the image processor 502. As explained above, the image processor 502 detects the worker in the cropped image. In this embodiment, in addition to detecting the worker, it is also possible to determine at which location in the booth (P1, P2, P3, P4, or P0) the worker is located. This is because the position (P1, P2, P3, P4, or P0) of the cropped image is known in advance from preset information.

[0098] Next, the extracted image (position P1, P2, P3, P4 or P0) data is stored in chronological order in the memory 503 for seated position data and staying time data.

[0099] Next, occupied position data indicating the occupied position in memory 503 and data on the stay time when the person stayed at that position are read out and processed by a conversion processor (also referred to as a time series conversion processor) 504 to convert the occupied position into occupied time series pattern data. Note that the occupied position is identified as occupied position data at one-minute intervals, as in the previous embodiment. The occupied position data is then stored in memory 503 according to the rules explained in FIG. 5.

[0100] The memory 503 is accessed by a time series conversion processor 504, which creates a time series pattern in which the three items P1, P2, and P3 are time-series according to their respective stay times (stay order), or a time series pattern in which the six items P1, P2, P3, P1, P2, and P3 are time-series according to their respective stay times (stay order).

[0101] The seating time series pattern data converted by the time series conversion processor 504 is stored in memory 505. Here, for example, 15-minute data in 15-minute units is transferred from the time series conversion processor 504 to the memory 505 and stored therein. In other words, data is transferred from the conversion processor 504 to the memory 505 every 15 minutes. This interval can also be set by preset information. This is to ensure a grace period for intensive processing such as the time series conversion processor 504 and the pattern data extraction processor 506.

[0102] The subsequent data processing is the same as that described in the case of Fig. 1, and pattern data extraction processor 506 compares the plurality of occupied time series pattern data from memory 505 with the good occupied time series pattern data provided from memory 507. Next, it is determined how many pieces of pattern data identical to the good occupied time series pattern data exist among the plurality of occupied time series pattern data (detected data) (determination of the number of occurrences of good occupied time series pattern data). This determination interval can also be set by preset information. Data indicating the number of occurrences is stored in memory 508.

[0103] The data indicating the occurrence number is given to a production number prediction data calculator 602 in the production number prediction unit 600. The production number prediction data calculator 602 is connected to a memory (not shown) having a linking table between the good attendance time series pattern data and the production number data of the product. As a result, the display unit displays the production number prediction data of the product for each set time period.

[0104] As described above, one embodiment is a production system in which a single booth has multiple work sites and a desired product is produced by moving between the multiple work sites. In such an environment, the number of completed products can be accurately predicted.

[0105] In another embodiment, in the production system of the above type, data on work sequences (good occupancy time series pattern data and poor productivity occupancy time series pattern data) that are efficient for work (and can increase the number of finished products) can be obtained. The occupancy time series pattern data is temporarily stored in memory, so by using this data, managers can obtain knowledge regarding the quality of work efficiency. In other words, the occupancy time series pattern data can be used for learning and training. Furthermore, it can also be used as data for devising work positions (occupancy positions).

[0106] Furthermore, in the production site of the above-mentioned type of production system, it is possible to record and manage the cycle of a single worker's seating position when he or she works at multiple work sites (for example, the first, second, third, and fourth work sites) and the worker's time spent at each seating position, and this data can be effectively used to improve production efficiency. For this purpose, a transfer device may be provided separately to transfer the data in the memory to a research processor.

[0107] While the above-described embodiment is an example of its application to production equipment, this work performance prediction system, equipment, and service provision system can be applied to any object that constitutes a work environment where work performance is expected. The work environment is not necessarily limited to production and manufacturing, but can also be applied to "processing," "treatment," and "repair" processes, such as dismantling waste such as equipment, repairing broken electrical appliances, and inspecting specimens. In other words, it is a simple invention for evaluating the efficiency of a series of work processes. (Work to dismantle waste) For example, in a work area where waste is dismantled, multiple work sites may be set up, and one worker may move to each work site to perform work.

[0108] For example, at a first work site, a worker selects several dismantling tools from a variety of options and heads to a second work site. At the second work site, the worker removes the second part (exterior case), the third part (motor), and the fourth part (circuit board). Next, the removed third part (motor) is taken to the third work site, the removed fourth part (circuit board) is taken to the fourth work site, the removed fifth part (base) is taken to the fifth work site, and then the fourth part at the fourth work site is further disassembled (removing wiring, etc.). A part recovery person comes to each work site to recover the disassembled parts.

[0109] In the above work, depending on the condition of the product being dismantled (deformation, corrosion, etc.), secondary tasks (repairing deformed parts, cleaning, etc.) may occur, especially at the first site (workers may return to the first site). This changes the time series pattern of time spent at the work site. When equipment with different designs by different manufacturers is randomly brought in, the time series pattern also changes. (For repairs of charged products) Even in repairing electrical appliances, multiple work sites are set up, and the following process can be considered: In the first work site, a search and inspection is carried out to find the faulty part of a certain device, and depending on the search results, The second work site may involve, for example, replacing a component on a circuit board, or At the third work site, replacement must be performed on a board-by-board basis.

[0110] Next, after the parts or boards are replaced, the product is returned to the first work site and inspected and checked again.

[0111] Typically, the process goes from the first work site to the second or third work site, and then the first work site goes through a routine, with the equipment finally being taken to the fourth work site after repairs are completed. However, when a final check is performed at the first work site, a failure or malfunction may be found again. For example, the initial failure may be fixed, but normal characteristics may not be achieved even after repairs. This will result in an increase in secondary work (additional work processes). Furthermore, there may be cases at the second or third work site where there is a shortage of parts or boards, or different parts or boards are needed. This will result in an increase in secondary work. (Examination of specimens) In sample testing, workers may move between multiple rooms or booths, i.e., multiple work sites, depending on the samples and chemicals used. While several embodiments of the present invention have been described, these embodiments are presented by way of example only and are not intended to limit the scope of the invention. These novel embodiments may be embodied in a variety of other forms, and various omissions, substitutions, and modifications may be made without departing from the spirit and scope of the invention. These embodiments and their modifications are within the scope and spirit of the invention, and are encompassed by the claims and their equivalents.

[0112] Furthermore, the scope of the present invention also includes cases in which each component of the claims is expressed separately, expressed as a plurality of components, or expressed as a combination of these. Furthermore, multiple embodiments may be combined, and examples formed by such combinations also fall within the scope of the invention.

[0113] In addition, the drawings may be more schematic than the actual embodiment for clarity of explanation, but these are merely examples and do not limit the interpretation of the present invention. Furthermore, in this specification and each drawing, components that perform the same or similar functions as those described above with reference to the previous drawings are designated by the same reference numerals, and redundant detailed descriptions may be omitted as appropriate. The device of the present invention is also applicable when the claims are expressed as control logic, a program including instructions for causing a computer to execute the program, or a computer-readable recording medium containing the instructions. Furthermore, the names and terms used are not limited, and other expressions that have substantially the same content and intent are also included in the present invention. [Explanation of symbols]

[0114] 200: Booth, 300: Camera, 400: Interface, 500: Time series pattern appearance detection unit, 501: Capture, 502: Image processor, 503: Memory for seat location data and stay time data, 504: Conversion processor to seat time series pattern data (time series conversion processor), 505: Memory for seat time series pattern data, 506: Extraction processor for excellent seat time series pattern data (pattern data extraction processor) device), 507...Memory of specific (good attendance time series) pattern data, 508...Memory of number of occurrences of good attendance time series pattern data, 600...Production number prediction unit, 601...Memory of association (linking) table between good attendance time series pattern data and production number data, 602...Production number prediction data calculator, 700...Warning device, 800...Display unit, 801...Operation unit, 900...Production quantity prediction table creator, 1111...System control unit.

Claims

1. A plurality of work positions are set within a predetermined work area, and the work environment is set so that a worker can move to the plurality of work positions and perform work to obtain a work result, The apparatus includes a storage unit, a pattern appearance detection unit, and a prediction unit, The storage unit stores occurrence count data of a good occupation time series pattern that has good performance data of the work result, the good occupation time series pattern being preset among a plurality of types of occupation time series patterns that are formed by the worker moving to a plurality of the work positions, and the work result data of the work result corresponding to the occurrence count data. and remember it. the pattern appearance detection unit acquires the appearance count data of the good occupancy time series pattern from the occupancy time series patterns detected when the work is actually performed, The prediction unit identifies the occurrence count data stored in the memory unit based on the occurrence count data acquired by the pattern occurrence detection unit, and obtains the result data corresponding to the identified occurrence count data as the work result.

2. 2. The work performance prediction system according to claim 1, further comprising a learning device that generates data of the good occupancy time series pattern and stores the data in the storage unit.

3. 2. The system for predicting work results according to claim 1, wherein the work areas are multiple.

4. The system for predicting work results according to claim 3 , wherein the prediction unit sums up the predicted production numbers for the plurality of work areas.

5. an interface for receiving an image signal from a camera that captures an image of the inside of the booth and providing the image signal to the pattern appearance detection unit; The interface is an image buffer memory for storing the image signal; a preset information storage unit capable of storing preset information used to extract extracted image signals corresponding to a plurality of seating positions in the booth from the image signal in the image buffer memory; The work performance prediction system of claim 1 , comprising:

6. A method for predicting a work outcome, which is applied to a work environment in which a plurality of work positions are set within a predetermined work area, and a work outcome is obtained by a worker moving to the plurality of work positions and performing work, comprising: A step of controlling a storage unit, a pattern appearance detection unit, and a prediction unit, a first step of storing in the storage unit occurrence count data of a good occupation time series pattern having good performance data of the work result corresponding to the occurrence count data, the good occupation time series pattern being preset among a plurality of types of occupation time series patterns formed by the worker moving to the occupation position; the pattern appearance detection unit includes a second step of acquiring the appearance count data of the good occupancy time series pattern from the occupancy time series patterns detected when the work is actually performed, The method for predicting work results comprises a third step in which the prediction unit identifies the occurrence count data stored in the memory unit based on the occurrence count data acquired by the pattern occurrence detection unit, and obtains the result data corresponding to the identified occurrence count data as work results.

7. A device for predicting work results, which is installed in a work environment in which a plurality of work positions are set within a predetermined work area, and a worker moves to the plurality of work positions to perform work, thereby obtaining work results, The apparatus includes a storage unit, a pattern appearance detection unit, and a prediction unit, The storage unit stores occurrence count data of a good occupation time series pattern that has good performance data of the work result, the good occupation time series pattern being preset among a plurality of types of occupation time series patterns that are formed by the worker moving to a plurality of the work positions, and the work result data of the work result corresponding to the occurrence count data. and remember it. the pattern appearance detection unit acquires the appearance count data of the good occupancy time series pattern from the occupancy time series patterns detected when the work is actually performed, The prediction unit identifies the occurrence count data stored in the memory unit based on the occurrence count data acquired by the pattern occurrence detection unit, and obtains the outcome data corresponding to the identified occurrence count data as the work outcome.

8. A plurality of work positions are set within a predetermined work area, and the system is installed in a work environment in which a worker moves to the plurality of work positions to perform work and obtains work results, and predicts the work results. A service providing system, a system control unit that controls the memory unit, the pattern appearance detection unit, and the prediction unit; the storage unit stores occurrence count data of a good occupation time series pattern, which is a pre-defined good occupation time series pattern having good performance data of the work result, among a plurality of types of occupation time series patterns formed by the worker moving to a plurality of the work positions, in association with the work result data of the work result corresponding to the occurrence count data; the pattern appearance detection unit acquires the appearance count data of the good occupancy time series pattern from the occupancy time series patterns detected when the work is actually performed, the prediction unit identifies the occurrence count data stored in the storage unit based on the occurrence count data acquired by the pattern occurrence detection unit, and obtains the result data corresponding to the identified occurrence count data as the work result; The service providing system is characterized in that the system control unit transmits data on the number of occurrences of the good occupancy time series patterns and the result data to an external device.

Images