KR101192592B1 - A flow-model statistical processing system for analyzing the manufacturing process - Google Patents

Abstract

A flow modeling based statistical processing system for generating a statistical processing graph composed of statistical processing unit operations, receiving data and processing respective unit operations of the graph to display a final processing result, comprising: a data obtaining unit receiving data; Generates statistical graph of data by receiving input / output relation between unit operation and unit operation, which is composed of node and directional edge, and node corresponds to unit operation A flow chart design unit generating a time point corresponding to an output of the unit task and an end point of the edge corresponding to an input of the unit task; A flow chart execution unit which processes the unit work of the graph in the order of the direction of the statistical processing graph, inputs and processes the output of the unit work as an input of the unit work according to the direction of the edge of the graph; And a result display unit for displaying the processing result.
By the system as described above, by performing the desired statistical analysis to combine the already created statistical processing unit work, it is possible to simply process complex statistical analysis with simple knowledge in the field.

Description

A flow-model statistical processing system for analyzing the manufacturing process}

The present invention relates to a flow modeling-based statistical processing system for generating a statistical processing graph consisting of statistical processing unit operations, receiving data and processing each unit operation of the statistical processing graph to display a final processing result.

In particular, the present invention relates to a flow modeling-based statistical processing system that can be modeled while directly displaying a continuous processing relationship between unit operations by having an input window of a user graphical interface (GUI).

In general, a lot of efforts are made to improve productivity at the production site, and the main system used to achieve this purpose is a static system such as Statistical Process Control (SPC). Such a system is installed at each major process to monitor and monitor the state of the process at all times.

These static systems are being developed and customized by professional IT companies when building production facilities. Therefore, only fixed values are displayed and managed, and the management point is managed with a fixed experience value. Also, in the management, the management by the station is mainly performed, and high-level analysis such as the analysis of the interrelationship between the processes and the scientific and statistical prediction of the production line due to the change of the main parameters cannot be performed.

According to the development of statistical analysis method, if various types of high-dimensional analysis methods are applied to the production site, it is possible to increase the possibility of predicting the production line according to the inter-process correlation analysis or parameter change. However, it is difficult to apply these various types of high-level analysis methods to the field.

In other words, in order to apply complex statistical analysis methods, it must be done through expert analysis. In addition, such an analysis task should be preceded by a task of extracting and transforming data suitable for statistical analysis processing. For this reason, there are many manpower problems or systemic problems that are difficult to apply to the production site where important decisions must be made every hour. Moreover, since various types of appropriate analysis algorithms have to be applied to various production systems, such customization is also difficult to perform at the production site.

Therefore, for a variety of more complex statistical analysis, there is an urgent need for a technique that can easily define a series of processes from data acquisition, analysis, and data output.

SUMMARY OF THE INVENTION An object of the present invention is to solve the problems described above, and generates a series of processes ranging from data acquisition, analysis, and output to a statistical processing graph composed of unit operations, and receives data for each unit operation of the graph. The present invention provides a flow modeling-based statistical processing system that processes and displays (outputs) final processing results.

It is also an object of the present invention to provide a flow modeling-based statistical processing system that can be modeled while directly displaying a continuous processing relationship between unit operations with an input window of a user graphical interface (GUI).

In order to achieve the above object, the present invention relates to a flow modeling-based statistical processing system that generates a statistical processing graph composed of statistical processing unit operations, receives data and processes each unit operation of the graph to display a final processing result. A data acquisition unit receiving the data; Generates a statistical processing graph of the data by receiving a statistical processing unit task having an input and an output and an input / output relationship between the unit operations, wherein the statistical processing graph comprises a node and a directional edge, and the node corresponds to the unit task. A flow chart design unit for generating a point of time of the directional edge corresponding to an output of a unit task and an end point of the edge corresponding to an input of a unit task; A flow chart execution unit which processes the unit operations of the graph in the order of direction of the statistical processing graph and inputs and processes the output of the unit operations as input of the unit operations according to the direction of the edge of the graph; And a result display unit for displaying the result of the processing.

In addition, the present invention, in the flow modeling-based statistical processing system, when the unit operation has two or more inputs or outputs, the unit operation node of the statistical processing graph is displayed on each input or output and the directional edge is one It corresponds to the input or output of.

In addition, the present invention is a flow modeling-based statistical processing system, the input or output of the unit operation is divided into input and output of a single value and input and output of an array (array), the starting point input and output and the endpoint input and output of the directional edge is the same format It is characterized by that.

In addition, the present invention is a flow modeling-based statistical processing system, characterized in that the size of the array is the same, if the starting point input and output of the directional edge is the input and output of the array type.

In another aspect, the present invention provides a flow modeling-based statistical processing system, the node starting from the statistical processing graph is a data input node, the unit operation of the data input node is to take the data file by the data acquisition unit to input data It is characterized by the operation.

In addition, the present invention is a flow modeling-based statistical processing system, the unit operation includes a display operation for displaying the input data in a statistical graph, when the unit operation is a display operation, there is no output of the unit operation do.

In another aspect of the present invention, in the flow modeling-based statistical processing system, the result display unit performs the display operation to display the input data of the display operation in a statistical graph.

In another aspect of the present invention, there is provided a flow modeling-based statistical processing system, wherein the flowchart design unit includes input means including an input window of a user graphic interface (GUI), wherein the input window includes a display window and a unit for displaying the statistical processing graph. And a job selection window for selecting a job, wherein the input means displays the unit job on the display window when receiving an input for moving the unit job selected from the job selection window to the display window, and connects the unit jobs in the display window. When the input is received, the directional edge is generated and displayed.

As described above, according to the flow modeling-based statistical processing system according to the present invention, for a variety of complex analysis, a series of processes ranging from data acquisition, analysis, output to simple implementation through the flow definition base, to various production sites The effect of easily implementing a suitable assay is obtained.

In addition, according to the flow modeling-based statistical processing system according to the present invention, it is possible to define the results obtained through a series of processes in the form of a report, outputs a variety of analysis results in the flow defined only without a separate complex programming process in the field A possible effect is obtained.

1 is a diagram showing an example of the overall system configuration for implementing the present invention.
2 is a block diagram of a configuration of a flow modeling-based statistical processing system according to an embodiment of the present invention.
3 is an example of a statistical processing graph created in accordance with the present invention.
4 is an example of a statistical processing graph in which inputs and outputs are displayed in accordance with the present invention.
5 is an example of a screen for a statistical processing graph design operation according to the present invention.
6 shows examples of statistical graphs output according to the present invention.

Description of the Related Art [0002]
10: manufacturing production system 11: work process
12: workflow 13: field data
14: data load unit 15: data set
20: network 21: administrator terminal
22: user terminal
30: statistical processing system 31: data acquisition unit
32: flow chart design unit 33: result display unit
34: flowchart execution unit
40: database 41: unit work DB
42: graph DB 50: input window
51: job selection window 52: display window

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the drawings.

In addition, in describing this invention, the same code | symbol is attached | subjected and the repeated description is abbreviate | omitted.

First, examples of the overall system configuration for implementing the present invention will be described with reference to FIG.

As shown in FIG. 1, the entire system for implementing the present invention is composed of a manufacturing production system 10, a data collection unit 15, a statistical processing system 30, and a manager or user terminal 21, 22. .

The manufacturing production system 10 refers to a manufacturing line or a production line that manufactures and produces a product at an industrial site. The manufacturing production process of the product is the final product is produced through a series of work processes (11). When one work process 11 is finished, the next work process 11 is performed. That is, one product is each work process 11 is performed according to the work flow 12, the final product is produced when all the work process 11 is finished.

On-site data 13 is collected for process control or quality control of the manufacturing production system 10. The site data 13 may be collected as measurement data in real time by installing various sensors or the like in the work process 11 or may be collected manually by a site worker.

The collected field data 13 is collected and stored in the data set 15 through the data load unit 14. The data load unit 14 transforms the data so as to extract or store the field data 13 to be stored.

The data set 15 is implemented as a data-ware house or a data mart. In other words, the collected data is stored in a standardized form, and when necessary data is required, the collected data can be extracted and used.

Statistical processing system 30 is a system that takes the data required from the data set 15, performs statistical analysis and outputs the result to a user or an analyst. The statistical processing system 30 defines a statistical processing flow (or statistical processing graph) composed of statistical processing unit operations, processes the unit operations according to the defined statistical processing flow, and outputs a final result.

Statistics processing units are predefined and modularized processing units modules. Therefore, the statistical processing flow can be defined only by the flows of the statistical processing unit operations already made, thereby defining the necessary statistical operations. On the other hand, when processing the unit work, the field data stored in the data set 15 is input and used.

Statistics flow is defined by the administrator and handled by the end user. The administrator and the user may be the same person. In this case, the manager or the user accesses the statistical processing system 30 through the manager terminal 21 or the user terminal 22 and performs necessary tasks. Therefore, hereinafter, the description of a manager or a user performing a task means a task performed by the manager terminal 21 or the user terminal 22. Incidentally, reference numerals 21 and 22 are also attached to the administrator or the user in addition to the terminal.

The manager terminal 21 or the user terminal 22 is a conventional computing terminal such as a PC, a notebook computer, a netbook, a PDA, a mobile device used by an administrator or a user.

The manager 21 defines (or designs) a statistical processing flow (or statistical processing graph) for the statistical analysis to be analyzed. Designing a statistical flow involves designing how the data will be analyzed. These tasks require basic knowledge of statistics and are therefore performed by administrators or analysts.

The general user 22 performs the desired statistical analysis by executing the generated statistical processing flow and outputs the result. At this time, the general user 22 inputs the field data stored in the data set 15 to perform the desired statistical analysis.

On the other hand, in order to provide such a service, the statistical processing system 30 is connected to the network 20 may be implemented as a web server or a web application server online. That is, a web page for providing a service to the manager terminal 21 or the user terminal 22 is provided, and the processed result is displayed on the web page or necessary input data is transmitted through the web page. The web page is meant to include driving software for performing a specific task, such as a web application, in addition to simple text, images, multimedia, and the like. Since such a technique is well known in the art, a detailed description thereof will be omitted.

The database 40 is a general storage medium for storing data necessary for the statistical processing system 30, and stores a unit operation DB 41 for storing unit modules capable of performing statistical processing, and a defined statistical processing graph. And a graph DB 42. The configuration of the database 40 is only one preferred embodiment, and in developing a specific device, the database 40 may be configured in a different structure by a database construction theory in consideration of the ease and efficiency of access and search.

Next, the configuration of the flow modeling-based statistical processing system 30 according to an embodiment of the present invention will be described in more detail with reference to FIG. 2.

As shown in FIG. 2, the statistical processing system 30 according to the present invention includes a data acquisition unit 31, a flowchart design unit 32, a flowchart execution unit 33, and a result display unit 34.

The data acquisition unit 31 receives data. The input data is the raw data for statistical analysis. As described above, data may be imported from the data set 15 such as a data warehouse or a data mart, or data may be directly input to a file by a user.

The flowchart design unit 32 receives a statistical processing unit task having an input and an output and an input / output relationship between the unit operations and generates a statistical processing graph of the data.

As shown in Fig. 3, the statistical processing graph is composed of nodes A1, A2, ..., A8, A9 and directional edges f13, f24, ..., f68, and the nodes correspond to unit operations. The start point of the directional edge corresponds to the output of the unit task and the end point of the edge corresponds to the input of the unit task.

In FIG. 3, nodes are represented by nine nodes (or unit operations), such as A1, A2, ..., A8, and A9. Nodes correspond to one unit of work, so use a combination of nodes or unit of work unless you need to distinguish them.

Each unit task is created as a series of flows beginning with A1 and ending with unit operations in A7 or A8. At this time, the directional edges indicating the flow are represented by f13, f24, ..., f67, f68. That is, fij and i denote a node number corresponding to the edge of the edge, and j denotes a node number corresponding to the edge of the edge. For example, the start and end points of edge f24 are nodes (or unit operations) A2 and A4, respectively.

The statistical processing graph of FIG. 3 uses the first data to train the neural network model to obtain coefficients of the neural network model, and to take the second data to be analyzed and apply the trained neural network model to a statistical processing operation for calculating prediction data. It is displayed. The statistical processing graph shows a statistical analysis processing task using a neural network model composed of a series of unit operations. The detailed description is as follows.

Unit tasks A1 and A2 are the unit tasks that start first. The first unit of work starts with no input. No input means no previous work. Further, preferably, the node to be started is a data input node, and the unit operation of the data input node is a job of importing a data file by the data acquisition unit 31 and inputting data.

A1 is a unit operation of opening an Excel file and acquiring data. Therefore, unit operation A1 defines information about an Excel file to be opened, a sheet on which data is located in the Excel file, or a column and a row. At this time, the data read from the Excel file in the unit operation A1 is a pair of an input value and an output value. That is, a set of pairs of input and output values for training a neural network model.

On the other hand, there is no input of the unit operation A1. The output of unit operation A1 is the set of pairs read from the Excel file. It can be seen that the output of A1 is a value of two array types. That is, an array for the input value input and an array for the output value output. Thus, the output of A1 has two outputs.

Unit operations A3 and A9 are operations for outputting standardized data by standardizing input data. Standardizing data is a unit of work that transforms data into a certain range or removes noise. For example, if the input data has a value in the range of 1 to 1000, the operation such as scaling the range of the data to -1.0 to 1.0 is described.

At this time, the unit operation A3 receives as input the input value input array which is one of the outputs of the unit operation A1, and the unit operation A9 receives as its input an output value output array which is the second of the outputs of the A1. Thus, the time point at edge f13 is the first output of A1, and the time point at f19 corresponds to the second output of A1.

Unit operations A3 and A9 are both unit operations that accept and standardize array values. As the manager 21 defines, the specific way of standardization may differ. For example, A3 may be scaled and output within a range of -1.0 to 1.0, but A9 may be output by being standardized into five selections (or five numbers 1,2,3,4,5). The inputs and outputs of unit operations A3 and A9 are one array value. That is, the number of inputs and outputs is one.

Unit task A5 is a unit task that learns by receiving a set of pairs (input, output). Thus the input has two array values. The first of the inputs of the unit operation A5 is input from the output of A3 and the second is input from A9. The edges representing each are f35 and f95. In other words, the start of edge f35 is the output of A3, and the end point of f35 is the first input of A5. The start of edge f95 is the output of A9 and the end of f95 is the second input of A5.

On the other hand, the output of unit operation A5 is the number and coefficients of hidden neural nodes. In this case, it is assumed that the hidden neural node is formed of only one layer. When the hidden neural node is composed of multiple layers, the format of the output data may vary accordingly.

Number of outputs Output is the number of hidden neural nodes, and coefficients during output are coefficients between each hidden neural node and an input value input or output value output. That is, the latter can be represented as an array. Thus, the output is two, the number consisting of a single value and the coefficient consisting of an array.

In the above example, the input or output of the unit operation is divided into a single value input and output and an array value input and output. Alternatively, data structures may be used in addition to single values and arrays. Since the data structure is a technology used in a data definition form used in a general programming language, a detailed description thereof will be omitted. In this embodiment, the output of the neural node may be defined as a structure, and when defined as a structure, only one output may be provided as a structure.

In the following description, it is assumed that the output of A5 has two outputs, a single value and an array value.

The unit work after the unit work A5 is A6, but in order for A6 to be performed, the previous work A2 and A4 must be performed.

Unit operation A2 is a task of obtaining data from a DB file. A2 is a task of inputting field data to obtain predictive data. Therefore, the data to be read in unit operation A2 is the value corresponding to input value in neural network. Therefore, data input from DB file can be one value or a series of values. The sequence of values is to get each of several pieces of data as an input, to get the output of each. If the number of values in the series is 1, it is a single value. Therefore, it is assumed that a series of values is input in the following description.

Therefore, the output of unit operation A2 is one array value.

Unit operation A4 is a standardization operation and is the same operation as unit operation A3. That is, since A3 and A4 are both standardized operations for input of neural networks, they must be processed in the same way.

Therefore, the start point of the edge f24 is one array value which is the output of the unit operation A2, and the end point is one array value which is the input of the unit operation A4. In other words, the data format of the start point and the end point of the edge f24 is the same.

Next, the output of the unit operation A4 is a standardized series of input values input, which is an array value. This value enters input of unit operation A6 via edge f46.

Unit operation A6 is a task of receiving neural network coefficients, receiving a series of input values, and outputting a series of output values corresponding to the series of input values. Thus, the input of unit operation A6 has a single value and two array values. The single value and one array value are received from unit operation A5 (this is received via edge 56), and the other array value is received from unit operation A4 (this is received via edge 46).

The output of unit operation A6 is a series of output values, one array value.

Unit operations A7 and A8 are the output of the output. That is, the unit operation of the statistical processing graph includes a display operation for displaying input data in a statistical graph, but there is no output of the unit operation when the unit operation is a display operation.

Unit work A7 is for printing as a table and A8 is for printing with a bar graph. At this time, the inputs of the unit operations A7 and A8 have one input, and the input is one array value. That is, it is followed by edges f68 and f67. In addition, the viewpoints of edge f67 and f68 are all one array value, and have the same output.

In addition, unit operations A7 and A8 have no output. Contrast that unit operations A1 and A2 have no input.

On the other hand, like unit operations A7 and A8, unit operations for displaying results may have various forms of output. The result data can be output in various forms such as a table, bar graph, and pie graph.

As shown in the example of FIG. 3, the statistical processing graph is composed of unit operations corresponding to nodes. At this time, the unit task uses the already created unit task module. That is, in the unit work module, like the template, the execution module and the form are predetermined, and the specific factors are determined by the manager 21. For example, unit operations A3, A4, and A9 use the same unit task module to perform standardization functions. However, the standard of A9 and the standard of A3 may be determined differently by the administrator.

On the other hand, in the statistical processing graph, when the unit task has two or more inputs or outputs, the unit task nodes display each input or output and the directional edges correspond to one input or output.

As shown in FIG. 4A, each unit task may be displayed in a square and the input or output of each unit task may be displayed in a small circle. In addition, the input or output of a single value is indicated by a general circle, and the input or output of the array value is indicated by a double circle. In the case of using a data structure, an image to be displayed may be changed according to each structure.

In addition, each directional edge connects one input at one output. In particular, two edges indicated by the unit operation A6 in the unit operation A5 are displayed. This is because both outputs of A5 are input to two inputs of A6.

In addition, the output of the start point of each edge and the input of the end point of the edge must be data of the same format. That is, the input or output of the unit operation is divided into input and output of a single value and input and output of an array value, and the start point input and output of the directional edge and the end point input and output are the same format.

Even in the case of using a data structure other than a single value and an array value in another embodiment, the structure of the start point and the end point of the edge must be the same.

In addition, as shown in Figure 4b, it is preferable to display each of the unit operations in the statistical processing graph as an icon imaged, so that the user or administrator can intuitively understand.

On the other hand, as shown in Figure 5, the flow chart design unit 32 is provided with an input means having an input window 50 of the user graphical interface (GUI).

In this case, the input window 50 is divided into a display window 52 for displaying a statistical processing graph and a job selection window 51 for selecting a unit job. In addition, the input unit displays the unit work on the display window 52 when receiving the input of moving the unit work selected in the job selection window 51 to the display window 52, and connects the unit jobs in the display window 52. When the input is received, the directional edge is generated and displayed.

The flowchart execution unit 33 processes the unit work of the graph in the order of the direction of the statistical processing graph, but inputs and processes the output of the unit work as an input of the unit work according to the direction of the edge of the graph.

That is, after the statistical process graph is defined in the flowchart design unit 32, the flowchart execution unit 33 performs each unit operation according to the defined statistical process graph.

In the example of FIG. 3, unit operations A1 and A2 which are start nodes are performed. The order of these is determined arbitrarily. That is, the unit operation is performed along the directional edge in the statistical graph. If the execution of the previous node is not completed in one node, the node is not performed. That is, in FIG. 3, unit operations A6 may be performed only when unit operations A5 and A4 are completed.

Also, in unit operations A1 and A2, if the flowchart design unit 32 defines only a file name, a file format to be input, or a data position in the file, the flowchart execution unit 33 inputs the actual data file. Do the work.

The unit task module of each unit task is a module defined as a function or a program. The flowchart execution unit 33 calls the module and executes the input argument of the module according to the input of the unit task or the defined content.

In addition, when one unit task is completed, the flowchart execution unit 33 inputs the output of the unit task to the input of the next unit task to perform the next unit task. At this time, the edge information of the graph is used.

In addition, the flowchart execution unit 33 verifies that the size of the array is also the same if the starting point input and output of the directional edge are the input and output of the array type. That is, the flowchart design unit 32 only defines the array format, but in practice, the array size at the output of the edge time point and the array size at the input of the edge end point should be the same. This is verified at run time and, if different, an error message is printed.

Finally, the result display unit 34 displays the processing result. The result display unit 34 performs a display operation (or a unit operation having a display function) to display the input data of the display operation in a statistical graph.

As shown in FIG. 6, a unit operation for outputting a graph may be output as various types of graphs. The result display unit 34 displays the result in a graph according to each unit operation.

As mentioned above, although the invention made by this inventor was demonstrated concretely according to the Example, this invention is not limited to an Example and can be variously changed in the range which does not deviate from the summary.

The present invention is useful for developing a flow modeling based statistical processing system that generates a statistical processing graph composed of statistical processing unit operations, receives data and processes each unit operation of the statistical processing graph to display a final processing result.

Claims (8)

In the flow modeling-based statistical processing system for generating a statistical processing graph composed of statistical processing unit operations, receiving data and processing the respective unit operations of the graph to display the final processing result,
A data obtaining unit which receives the data;
Generates a statistical processing graph of the data by receiving a statistical processing unit task having an input and an output and an input / output relationship between the unit operations, wherein the statistical processing graph comprises a node and a directional edge, and the node corresponds to the unit task. A flow chart design unit for generating a point of time of the directional edge corresponding to an output of a unit task and an end point of the edge corresponding to an input of a unit task;
A flow chart execution unit for processing the unit work of the graph in the order of the direction of the statistical processing graph, and inputting and processing the output of the unit work as an input of the unit work according to the direction of the edge of the graph; And,
A result display unit for displaying the result of the processing;
When the unit job has two or more inputs or outputs, the unit job node of the statistical graph displays all the inputs or outputs, and the directional edge corresponds to one input or output,
The input or output of the unit operation is divided into input and output of a single value and input and output of an array value, the starting point input and output of the directional edge is the same type,
If the starting point input and output endpoints of the directional edge is the input and output of the array type, the flow modeling based statistical processing system, characterized in that the size of the array.
delete delete delete The method of claim 1,
The node starting from the statistical processing graph is a data input node, and the unit operation of the data input node is a task of importing a data file and inputting data by the data acquisition unit.
The method of claim 1,
The unit operation includes a display operation for displaying the input data in a statistical graph, if the unit operation is a display operation, the flow modeling based statistical processing system, characterized in that there is no output of the unit operation.
The method of claim 6,
And the result display unit displays the input data of the display operation as a statistical graph by performing the display operation.
The method of claim 1,
The flowchart design unit may include input means including an input window of a user graphic interface (GUI), the input window including a display window for displaying the statistical processing graph and a job selection window for selecting a unit job,
The input unit displays the unit work on the display window when receiving the input of moving the unit work selected from the work selection window to the display window, and generates and displays the directional edge when receiving the input connecting the unit work on the display window. Flow modeling based statistical processing system, characterized in that.

Images