JP2017126164A - Data analysis device and data analysis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To complement production result data including pieces missing or errors.SOLUTION: An data analysis device includes a computer having: a processor configured to execute a program; and a storage unit storing the program. The storage unit includes: a data complement processing part configured to store manufacturing result data including a process sequence, a work type, identification information of an operated machine, work start time and work end time, and to complement missing data using data determined so as not to contradict with manufacturing result data of prior and later processes; and a data replacement processing part configured to replace replaceable data.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a data analysis apparatus, and more particularly to a technique for complementing manufacturing performance data.

  In the manufacturing industry, customer needs for delivering products with short delivery times are increasing. In addition, a short delivery time is preferable from the viewpoint of reducing the storage location of work in progress and improving the cash flow. For this reason, it is necessary to investigate which process of which order is a bottleneck, and to examine whether there is room for improvement in delivery time.

  These studies are actually based on manufacturing performance data indicating how the product was manufactured. In many cases, manufacturing performance data is acquired in a format as shown in FIG. 1 by a system called Point of Production (POP). For example, in the manufacturing performance data, for each process work, the order No100 indicating which customer order is associated with the work, the process number 101 indicating the order of the work in the order, the work type 102, and the work were performed. The ID 103 of the machine, the ID 104 of the person in charge who performed the work, the time 105 at which the work was started, and the time 106 at which the work was finished are recorded. In the example shown in FIG. 1, an order with an order No. 1 is divided into 10 processes, and the first work is “injection” performed by the person in charge “person-01” using the machine “injection-01”. This is a type of work, indicating that it started at 10:55 and ended at 11:45. Note that either the machine ID 103 or the person in charge ID 104 may be used. In other words, the machine ID 103 is unnecessary for work without using a machine, and the person in charge ID 104 is not required for automated work.

  However, in practice, it is difficult to obtain complete manufacturing performance data as shown in FIG. Each item of the manufacturing result data shown in FIG. 1 may be automatically registered by an industrial machine or automatically registered through RFID or the like, but may be manually registered by an operator during work. In addition, for the purpose of highly efficient and flexible manufacturing, work procedures, machines used, and work personnel may be adjusted as needed at the discretion of the manager at the work site. In such a case, the input of manufacturing frequency data that is excessively frequent and fine-grained causes interruption of the manufacturing work and causes a decrease in work quality and efficiency. It is difficult.

  For the reasons described above, part of the actual production data actually obtained is missing or incorrect.

  As a calculation process for data including a deficiency or an error, as shown in Non-Patent Document 1, a method has been proposed in which a deficit value is complemented by a substitution of a value calculated by a representative value, a sample average value, regression, or the like. (Method 1). Further, as shown in Patent Documents 1 and 2, a method for obtaining a distribution of a passing process pattern and a required time (manufacturing period) of the entire order has been proposed (Method 2).

Japanese Patent Laid-Open No. 2004-30088 JP 2010-128679 A JP 2013-33450 A

R. R. A. Little and D. B. Rubin (2002), Statistical analysis with missing data, Wiley

  However, it is difficult to accurately compensate for missing or incorrect manufacturing performance data by the method 1 described above. This is because substitution of missing values is effective for continuous values such as the time required for each operation, but it is difficult to supplement values such as machine ID, person in charge ID, start time, and end time. is there. In addition, in the method of performing calculation processing on the parameter model obtained by the maximum likelihood method or the Bayes method, or the method of obtaining the distribution of the required time of the passing process pattern and the entire order (method 2), the manufacturing period and each work are A parameter model of the time required can be obtained, but the true value of the missing or incorrect item cannot be obtained. For example, when complementation is performed using a predetermined value such as an expected value, a consistent description of actual manufacturing results is not guaranteed (for example, one machine has performed a plurality of operations at the same time). Further, by using a parameter model of the time required for each work, it is possible to examine a work that requires a relatively long work time. However, it is not possible to determine the contradiction between the work of the machine and the person in charge at each time point. In addition, in the method in which only the part that does not include missing values is subject to calculation processing, only a part of the manufacturing process data of the order in which only a part of the process constituting the order can be used or inconsistencies in the machine or the person in charge are generated. It is not possible to use it, and the examination result may not reflect the reality.

  This invention is made | formed in view of such a condition, and provides the technique which complements the manufacture performance data containing a defect | deletion and an error.

  A typical example of the invention disclosed in the present application is as follows. In other words, the data analysis apparatus is configured by a computer having a processor that executes a program and a storage device that stores the program, and the storage device includes process order, type of work, and identification information of a machine that has worked. Replaces replaceable data with data supplement processing unit that stores manufacturing result data including work start time and work end time, and complements missing data with data determined not to be inconsistent with manufacturing result data of previous and subsequent processes And a data replacement processing unit.

According to a typical embodiment of the present invention, manufacturing performance data including defects and errors can be supplemented. Problems, configurations, and effects other than those described above will become apparent from the description of the following embodiments.

It is a figure which shows the example of the manufacture performance data used in the Example of this invention. It is a functional block diagram which shows the structure of the data analyzer of a present Example. It is a figure which shows the data structure of the manufacture performance data of a present Example, exchangeable data, and order exchangeable data. It is a figure which shows the data structure of the process order relation data of a present Example. It is a flowchart which shows roughly the flow of the process performed in the data analyzer of a present Example. It is a flowchart which shows the detail of the missing data complementation process of a present Example. It is a flowchart which shows the detail of the maximum likelihood solution calculation process of a present Example. It is a flowchart which shows the detail of the early / late setting process of a present Example. It is a flowchart which shows the detail of the initial value setting process of a present Example. It is a flowchart which shows the detail of the exchangeable data replacement process of a present Example.

<Example 1>
Embodiments of the present invention will be described below with reference to the drawings.

  Hereinafter, a business document processing apparatus according to an embodiment of the present invention will be described with reference to the accompanying drawings. An embodiment described later is an example of the present invention, and the present invention includes various modifications.

  2 to 10 are diagrams illustrating examples described below. In these drawings, the same reference numerals denote the same components, the basic configuration is the same, and the operations thereof are the same. Is the same.

  FIG. 2 is a functional block diagram showing the configuration of the data analysis apparatus according to this embodiment.

  The data analysis apparatus includes a display device 200, a keyboard 201, a pointing device 202 such as a mouse, a central processing unit 203, a program memory 204, a data memory 205, a manufacturing performance database 206, and master data 207. . The display device 200 displays data output from the data analysis device. The keyboard 201 is operated to select a menu on the display screen. The central processing unit 203 executes arithmetic processing, control processing, and the like necessary for realizing the functions of the data analysis device. The program memory 204 stores a program executed by the central processing unit 203. The data memory 205 stores data necessary for processing executed by the central processing unit 203.

  The data analysis apparatus has a large-capacity and nonvolatile auxiliary storage device such as a magnetic storage device (HDD) and a flash memory (SSD), and the manufacturing performance database 206 and the master data 207 are stored in the auxiliary storage device. The program executed by the central processing unit 203 may be read from the auxiliary storage device, loaded into the program memory 204, and executed by the central processing unit 203.

  The central processing unit 203 includes a missing data supplement processing unit 208 and a replaceable data replacement processing unit 209. The data analysis apparatus of this embodiment is configured by a computer, and each of the processing units 208 and 209 is realized as part of the function of a program executed by the computer. The missing data complementing processing unit 208 includes a parameter model estimation processing unit 210 and a maximum likelihood solution calculation processing unit 211. The maximum likelihood calculation processing unit 211 includes an early / late limit value setting processing unit 212 and an initial value setting processing unit 213.

  The data memory 205 stores manufacturing performance data 214, exchangeable data 215, order exchangeable data 216, and process order relation data 217.

  The exchangeable data 215 and the order exchangeable data 216 are also stored in the master data 207.

  The program executed by the central processing unit 203 is provided to the data analysis apparatus via a removable medium (CD-ROM, flash memory, etc.) or a network, and is a non-temporary storage medium, a nonvolatile program memory 204 (or an auxiliary program) Stored in a storage device). For this reason, the data analysis device may have an interface for reading data from a removable medium.

  A data analysis device is a computer system that is configured on a single computer or a plurality of computers that are logically or physically configured, and operates on separate threads on the same computer. Alternatively, it may operate on a virtual machine constructed on a plurality of physical computer resources. In addition, each functional unit of the data analysis apparatus may be realized on a different computer.

  FIG. 3 is a diagram showing a data structure of manufacturing performance data 214, exchangeable data 215, and order exchangeable data 216 stored in the data memory 205.

  3 and 4, the data is described in a “table” format. However, these data do not necessarily have to be represented by a table data structure, such as a matrix, a list, a database, a queue, and the like. It may be expressed in other ways.

  As shown in FIG. 3A, the manufacturing result data 214 includes an order No. 300, a process order 301, a work type 302, a machine ID 303, a person-in-charge ID 304, a start time 305, and an end time 306. Held in. Each element of the array represents each row of manufacturing results shown in FIG. The order No. 300 is identification information for uniquely identifying the order. The process order 301 is an order of operations in the processes constituting the order. The work type 302 is the type of work. The machine ID 303 is identification information of the machine that performed the work. The person-in-charge ID 304 is identification information of the person in charge who performed the work. The start time 305 and the end time 306 are the time when the work is started and the time when it is finished, respectively. Note that either the machine ID 303 or the person-in-charge ID 304 may be one as in the production result data shown in FIG. That is, the machine ID 303 is not necessary for work that does not use a machine, and the person-in-charge ID 304 is not required for automated work.

  As shown in FIG. 3B, the exchangeable data 215 includes a column name 307, an exchangeable list 308, and a probability 309, and each item is held in the form of an array. The column name 307 is identification information of the column name. The exchangeable list 308 is a list of exchangeable values in the column. The probability 309 includes the probability that the above-described value exchange occurs, and is held in the form of an array. In the example shown in FIG. 3B, the work described in the “machine ID” column with “injection-01” in the manufacturing results shown in FIG. 1 can also be performed by the “injection-03” machine. The probability that the content recorded as the result and the work was performed by exchanging the machine is the probability 0.25.

  As shown in FIG. 3 (C), the order exchangeable data 216 includes two operations that can be performed in the manufacturing order shown in FIG. 311 and the occurrence probability is recorded in the occurrence probability 312. In the example shown in FIG. 3C, in the case where the order No. is x and the process order is y and y + 2, both perform “injection” operations on the machine “injection-02”. The processes can be performed in the reverse order, and the probability that the work was performed by exchanging the order with the contents recorded as the manufacturing results is 0.1. Note that, as described in step 500 of FIG. 5, since setup and conveyance are inserted as virtual work, two processes including one work in between are to be replaced.

  The exchangeable data 215 and the order exchangeable data 216 are specified by the physical properties of the machine and the work without depending on the manufacturing performance data 214, prepared in advance, and held in the master data 207.

  FIG. 4 is a diagram showing a data structure of the process order relation data 217 stored in the data memory 205.

  The process order relation data 217 includes Idx400, order No. 401, process order 402, start or end 403, front relation 404, rear relation 405, time 406, early limit 407 and late limit 408, and each item is held in the form of an array. Is done. Idx400 is the element number of the relationship. Order No. 401 is identification information for uniquely identifying an order. The process order 402 is a work order in the processes constituting the order. The start or end 403 represents either the start or end of the work. The pre-relationship 404 is an Idx of work performed before the work. The post relationship 405 is an Idx of work performed after the work. The pre-relationship 404 and the post-relationship 405 are held in the form of an array containing two elements, the first element indicates the work represented by the order relationship of the steps making up the order, and the second element is the same machine The operation | movement represented by the order relationship of the process which uses is shown. If it is the first or last step in the order, the first element of the pre-relationship 404 or the post-relationship 405 is NULL. If this is the first or last operation using the machine, the second element of the pre-relationship 404 or the post-relationship 405 will be NULL.

  The time 406 is the time designated by the start or end 403 of the work. The early limit 407 is a limit value indicating the earliest time value that can be selected. The delay limit 408 is a limit value that indicates what value is the latest and can be selected. That is, if the time 406 is set between the early limit 407 and the late limit 408, it is possible to obtain a solution with no time contradiction.

  Next, processing performed in the business document processing apparatus of the present embodiment configured as described above will be described.

  FIG. 5 is a flowchart schematically showing a flow of processing executed in the data analysis apparatus.

  In FIG. 5, first, the data held in the manufacturing record database 206 and the master data 207 are read into the manufacturing record data 214, the exchangeable data 215, and the order exchangeable data 216 (step 500). At this time, if the start time 305 and the end time 306 of the manufacturing result data 214 are the same value, or if the prior relationship is reversed, it is determined that the input value is incorrect, and the data is changed to NULL. To do. Also, setup and transfer operations are inserted one by one between the two processes.

  Next, the evaluation value is initialized (step 501). At this time, the evaluation value is a state in which errors and defects included in the manufacturing data are not complemented, and a sufficiently bad evaluation is assumed as a provisional value.

  Thereafter, the missing data is complemented (step 502). The processing of step 502 is executed by the missing data complementing processing unit 208, and details thereof will be described with reference to FIG.

  Thereafter, it is determined whether the evaluation value has converged (step 503). If it is determined that a better evaluation value cannot be obtained for a predetermined number of times or more, it is determined that it has converged, and the process is terminated. If an evaluation value better than the previous time is obtained, replaceable data is replaced (step 504). Specifically, the replaceable data replacement processing unit 209 performs a process corresponding to error compensation. Details of the processing in step 504 will be described with reference to FIG.

  FIG. 6 is a flowchart showing details of the missing data supplement processing in step 502 of FIG.

  First, the evaluation value is initialized (step 600). At this point in time, the evaluation value is a state in which the deficiency included in the manufacturing data is not supplemented, and a sufficiently bad evaluation is set as a provisional value.

  Next, the machine order is set (step 601). This process uses a restriction on the number of tasks that one machine can perform at the same time (for example, one machine can only perform one task at a time), and determines in which order each machine has worked. The result is held as the second element of the pre-relationship 404 and the second element of the post-relationship 405 of the process order relation data 217. If the start or end 403 is “start”, it is determined whether the start time 305 of the manufacturing performance data 214 is NULL. If not, the manufacturing performance is recorded at the time 406, the early limit 407 and the late limit 408. The start time 305 of the data 214 is set. On the other hand, if the start or end 403 is “end”, it is determined whether the end time 306 is NULL. If it is not NULL, the production result data 214 is ended at the time 406, the early limit 407, and the late limit 408. Time 306 is set.

  Thereafter, it is determined whether or not there is one or more solutions without contradiction (step 602). Specifically, the pre-relationship 404 and the post-relationship 405 of the process order relation data 217 are input relations of intermediate work in progress, and the time 406 is treated as material arrival and delivery time, and the delivery date is satisfied with the existing scheduler technology. If the schedule can be calculated, it is determined that there is no contradiction.

  Thereafter, a parameter model is calculated (step 603). In step 603, the parameter model inference processing unit 210 obtains a statistical value (for example, average or variance) of a time distribution required for work including setup and conveyance for a set of work types of two consecutive processes. . The average and variance of the time required for the work can be obtained by, for example, the method described in Patent Document 3.

  Thereafter, the maximum likelihood solution is calculated (step 604). In step 604, the maximum likelihood solution processing unit 211 uses a value determined to maximize the likelihood based on the acquired time, the assumed machine order, and the estimated parameter model. To complement. Details of the processing in step 604 will be described with reference to FIG.

  FIG. 7 is a flowchart showing details of the maximum likelihood calculation processing in step 604 of FIG.

  First, an early limit 407 and a late limit 408 are set for each element of the process order relation data 217 (step 700). The processing of step 700 is executed by the early / late limit value setting processing unit 212, and details thereof will be described with reference to FIG.

  Next, it is determined whether the subsequent processing has been repeatedly executed for a preset number of times (step 701). If it has been repeatedly executed a predetermined number of times or more, the maximum likelihood solution calculation process is terminated. If it has not been repeatedly executed a predetermined number of times, the subsequent processing is executed.

  Thereafter, an initial value is set (step 702). Specifically, the initial value setting processing unit 213 complements the missing value so that there is no randomness and no contradiction. Details of the processing in step 702 will be described with reference to FIG.

  Thereafter, the initial value set in step 702 is set as the current value (step 703), and the solution is evaluated (step 704). Specifically, using the parameter model obtained in step 603, the P value is obtained by the Kolmogorov-Smirnov test or the Shapiro-Wilk test, the P value is calculated for each type of work, and the number of replacements performed in step 504 The likelihood is obtained by calculating the power of the occurrence probabilities 309 and 312 according to the above and multiplying the power of the occurrence probability by the P value.

  Thereafter, it is determined whether the evaluation value has converged (step 705). If it is determined that the evaluation value better than the previous time cannot be obtained in step 704 for a predetermined number of times or more, it is determined that the evaluation value has converged, the process returns to step 702, and the process is repeated. On the other hand, if a better evaluation value than the previous time is obtained, or if the number of times that a better evaluation value cannot be obtained is less than the predetermined number, the processing after step 706 is executed.

  Thereafter, the supplementary content of each missing value is slightly increased or decreased from the current value set in step 703 to move the missing value (step 706), and it is determined whether there is a consistent solution even after the increase or decrease (step 707). . The processing in step 707 can be performed in the same manner as in step 602. If there is no consistent solution, the process returns to step 706 to move the stored missing value again. If there is a solution with no contradiction, the solution is evaluated in the same manner as in step 704 (step 708), and the evaluated solution is updated as a new current value (step 709).

  FIG. 8 is a flowchart showing details of the early / late setting process in step 700 of FIG.

  First, it is determined whether elements of the process order relation data 217 still exist (step 800). If the element of the process order relation data 217 still exists, it is determined whether the value of the early limit 407 is NULL (step 801). When the value of the early limit 407 is NULL, the latest value among the values of the time 406 and the early limit 407 that are not NULL in the element of the process order relation data 217 having the first or second element of the previous relation 404 as Idx400 Is used to update the value of early limit 407 (step 802).

  On the other hand, if the value of the early limit 407 is not NULL in step 801, it is determined whether the value of the late limit 408 is NULL (step 803). When the value of the delay 408 is NULL, the earliest value among the values of the time 406 and the delay 408 that are not NULL in the element of the process order relation data 217 having the first or second element of the post relation 405 as Idx400 Is used to update the value of the delay 408 (step 804) and the process returns to step 800.

  On the other hand, if the value of the delay 408 is not NULL in step 803, step 804 is skipped and the process returns to step 800.

  In step 800, if there is no unprocessed process order relation data 217, it is determined whether the early limit 407 or the late limit 408 has been updated in the repetition from step 801 to 804 (step 805). If at least one of the early limit 407 and the late limit 408 has been updated, the process is restarted from the first process order relation data 217 (step 806). If both the early limit 407 and the late limit 408 have not been updated, the early / late setting process is terminated.

  FIG. 9 is a flowchart showing details of the initial value setting process in step 702 of FIG.

  First, it is determined whether or not the process order relation data 217 whose time 406 is NULL still exists (step 900). If there is process order relation data 217 whose time 406 is NULL, one such process order relation data is selected, and a value between the early limit 407 and the late limit 408 is set to the time 406 (step 901). On the other hand, if there is no process order relation data 217 whose time 406 is NULL, the initial value setting process is terminated.

  Next, the value of the delay 408 is updated with reference to the previous relation 404 of the process order relation data 217 (step 902). When the value of the time 406 of the process order relation data 217 having the first and second elements of the previous relation 404 as Idx400 is not NULL, the value of the delay 408 is updated with the value selected in step 901. Further, the same processing is performed retroactively to the element of the process order relation data 217 having the first and second elements of the previous relation 404 of the element as Idx400.

  Next, the value of the early limit 407 is updated with reference to the post relation 405 of the process order relation data 217 (step 903). If the value of the time 406 of the process order relation data 217 having the first and second elements of the post relation 405 as Idx400 is not NULL, the value of the early limit 407 is updated with the value selected in step 901. In addition, the same processing is performed retroactively to the element of the process order relation data 217 having the first and second elements of the post relation 405 of the element as Idx400.

  Then, it returns to step 900 and repeats processing.

  FIG. 10 is a flowchart showing details of the replaceable data replacement processing in step 504 of FIG.

  First, it is determined whether there is a cell that has not yet been selected for replacement (step 1000). Specifically, it is determined whether there is an element that is not selected among the work type 302, the machine ID 303, and the person-in-charge ID 304 in the manufacturing performance data 214. If no element is selected, the process proceeds to step 1004.

  Next, one cell that has not yet been selected for replacement is selected (step 1001). Specifically, one of the cells determined not to be selected in step 1000 is selected.

  Thereafter, it is determined whether the cell selected in step 1001 is included in the exchangeable data 215 (step 1002). Specifically, the column name of the cell selected in step 1001 matches the column name 307 of the replaceable data 215, and the value of the cell selected in step 1001 is in the replaceable list 308 of the replaceable data 215. Determine if it is included in the element. As a result, if the selected cell is not included in the exchangeable data 215, the process returns to step 1000 and the process is repeated. If the selected cell is included in the exchangeable data 215, the data is exchanged (step 1003). Specifically, the value of the cell is updated with one value selected from the exchangeable list of exchangeable data 215.

  Thereafter, it is determined whether there is a row that has not yet been selected as an exchange target (step 1004). Specifically, it is determined whether there is an array element that has not yet been selected in the manufacturing performance data 214. If all the rows have been selected as exchange targets, the exchangeable data replacement process is terminated.

  Thereafter, one row is selected from the rows that have not yet been selected as exchange targets (step 1005). Specifically, one is selected from the rows determined not to be selected in step 1004.

  Thereafter, it is determined whether or not the row selected in step 1005 is included in the first row 310 of the order exchangeable data 216 (step 1006). Specifically, it is determined whether the value of the array element selected in step 1005 matches the contents of the first row 310 of any array element of the order exchangeable data 216. If the value of the selected array element does not match the contents of the first row 310 of another array element, the process returns to step 1004.

  On the other hand, if it is determined that the value of the array element selected in step 1005 matches the content of the first row 310 of another array element, the row selected in step 1005 becomes the first row of the order exchangeable data 216. It is determined whether it is included (step 1007). Specifically, it is determined whether the array element at the relative position specified in the second row 311 of the array element selected in step 1005 matches the contents of the second row 311. If the array element at the relative position specified by the second row 311 of the selected array element does not match the contents of the second row 311, the process returns to step 1004.

  On the other hand, if the array element at the relative position specified in the second row 311 of the array element selected in step 1005 matches the contents of the second row 311, the data is exchanged (step 1008). Specifically, this is an operation of replacing values other than the process order 301 between the array element selected in step 1005 and the array element at the relative position specified by the second row 311 of the array element.

  As described above, according to the embodiment of the present invention, manufacturing performance data including defects and errors can be appropriately supplemented. As a result, it is possible to investigate which process of which order is a bottleneck and to examine whether there is room for improvement in delivery time.

  In addition, since the missing data complement processing unit 208 supplements the missing data with the data determined to maximize the likelihood calculated based on the determined work order and the estimated parameter model, Data can be complemented.

  In addition, since the missing data complement processing unit 208 (maximum likelihood calculation processing unit 211) sets the start time 305 and the end time 306 in consideration of the early limit 407 and the late limit 408, there is a contradiction in relation to other processes. You can choose a solution without

  In addition, since the exchangeable data 215 and the order exchangeable data 216 are referred to determine whether or not the exchange is possible, the data can be corrected to reflect the reality.

  The present invention is not limited to the above-described embodiments, and includes various modifications and equivalent configurations within the scope of the appended claims. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and the present invention is not necessarily limited to those having all the configurations described. A part of the configuration of one embodiment may be replaced with the configuration of another embodiment. Moreover, you may add the structure of another Example to the structure of a certain Example. In addition, for a part of the configuration of each embodiment, another configuration may be added, deleted, or replaced.

  In addition, each of the above-described configurations, functions, processing units, processing means, etc. may be realized in hardware by designing a part or all of them, for example, with an integrated circuit, and the processor realizes each function. It may be realized by software by interpreting and executing the program to be executed.

  Information such as programs, tables, and files that realize each function can be stored in a storage device such as a memory, a hard disk, and an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, and a DVD.

  Further, the control lines and the information lines are those that are considered necessary for the explanation, and not all the control lines and the information lines that are necessary for the mounting are shown. In practice, it can be considered that almost all the components are connected to each other.

200 ... Display device 201 ... Keyboard 202 ... Pointing device 203 ... Central processing unit 204 ... Program memory 205 ... Data memory 206 ... Manufacturing performance database 207 ... Master data

Claims (8)

A data analysis device,
A computer having a processor for executing the program and a storage device for storing the program;
The storage device stores production result data including process order, work type, machine identification information, work start time and work end time,
A data supplement processing unit that supplements missing data with data determined so as not to contradict manufacturing performance data of previous and subsequent processes;
A data analysis apparatus comprising: a data replacement processing unit that replaces exchangeable data. The data analysis device according to claim 1,
The data complement processing unit
Using the restriction on the number of tasks that one machine can perform at the same time, determine the work order of each machine,
Estimate the parameter model by obtaining the statistics of the time required for the work between two consecutive processes,
The missing data is supplemented with data determined to maximize the likelihood calculated based on the time when the work was performed, the determined work order, and the estimated parameter model. Data analysis device. The data analysis device according to claim 2,
The data complement processing unit
Set the start time of each step in consideration of the earliest and latest limit values,
The end time of each step is set in consideration of the earliest and latest limit values,
A data analysis apparatus characterized in that the set start time and end time determine data having no contradiction in relation to other processes. The data analysis device according to claim 1,
The storage device stores exchangeable data that defines exchangeable data and order exchangeable data that defines exchangeable data.
The data replacement processing unit
With reference to the exchangeable data, the type of work of the manufacturing performance data and the identification information of the machine that has worked are determined to be exchangeable,
With reference to the order exchangeable data, it is determined whether the process of the manufacturing performance data is exchangeable,
A data analysis apparatus that replaces the data determined to be exchangeable. A data analysis method in which a computer analyzes manufacturing performance data,
The computer includes a processor that executes a program, and a storage device that stores the program.
The storage device stores production result data including process order, work type, machine identification information, work start time and work end time,
The data analysis method includes:
A data complementing step for complementing missing data with data determined so that the processor is consistent with manufacturing performance data of previous and subsequent processes;
A data analysis method, wherein the processor includes a data replacement processing step of replacing replaceable data. The data analysis method according to claim 5, comprising:
In the data complementing process step,
The processor determines the work order of each machine using a restriction on the number of work that one machine can perform simultaneously,
The processor infers a parameter model by obtaining a statistic of the time required for work between two successive steps,
The processor supplements missing data with data determined to maximize likelihood based on the time when the work was performed, the determined work order, and the estimated parameter model. Data analysis method. The data analysis method according to claim 6, comprising:
In the data complementing process step,
The processor sets the start time of each step in consideration of the earliest and latest limit values,
The processor sets the end time of each step in consideration of the earliest and latest limit values;
The data analysis method, wherein the processor determines data in which the set start time and end time are consistent with each other in relation to other processes. The data analysis method according to claim 5, comprising:
The storage device stores exchangeable data that defines exchangeable data, and order exchangeable data that defines exchangeable data in order,
In the data replacement processing step,
The processor refers to the exchangeable data, and determines whether or not the type of work of the manufacturing performance data and the identification information of the machine that has worked are exchangeable,
The processor refers to the order exchangeable data, determines whether or not the manufacturing performance data process is exchangeable,
The data analysis method, wherein the processor replaces data determined to be exchangeable.

Images