KR20200002433A - Statistical quality control system and method using big data analysis - Google Patents

Abstract

The present invention relates to a statistical quality management system using big data analysis and a method thereof. The statistical quality management system includes: a collecting unit which indexes analysis target data, learning target data, and a learning model from a collecting target equipment to load the same to a storage; an analyzing unit which analyzes that which learning model is suitable for the analysis target data and the learning target data to generate analysis information; a learning unit which matches the analysis target data and the learning target data with a learning model according to the analysis information to perform deep learning; a maintenance history management unit which quantifies maintenance history for each of analysis target data and learning target data corresponding to the collecting target equipment based on a result of deep learning; and a detection complementing unit which analyzes connection for causal relationship, precision, error, and defective state for the collecting target equipment and the maintenance history to measure reliability and precision. According to the present invention, provided is an effect of providing analysis on a numerical value of generation and an estimated value for error detection, cause of error, error generated line, staff information, and process information.

Description

Statistical quality control system and method using big data analysis

The present invention relates to a statistical quality management system and method using big data analysis, and more particularly, to distribute the learning for the classification, analysis, prediction and diagnosis of big data collected from plant control / monitoring / production facilities However, the present invention relates to a checklist for 4M (Man / Machine / Material / Method) analysis and a technique for performing learning for 5 way analysis.

Big data refers to a large set of structured or unstructured data sets that go beyond the ability to collect, store, manage, and analyze data with traditional database management tools, and the technology to extract value from these data and analyze the results.

The development of big data technology, which is characterized by the creation, collection, analysis, and expression of various kinds of large-capacity data, enables more accurate prediction of diversified modern society to operate efficiently, and provides customized information for each personalized modern society member. Allow management and analysis.

As such, big data shows the possibility of providing useful information to society and mankind in all areas such as politics, society, economy, culture, science and technology, and its importance is highlighted.

Big data analysis can involve data mining, machine learning, natural language processing, and pattern recognition, which were used in traditional statistics and computing. In particular, text mining, opinion mining, social network analysis, cluster analysis, and the like have attracted attention due to the increase in unstructured data such as social media.

However, the conventional big data collection has a disadvantage in that the logic of the program needs to be reconfigured or re-developed when a parameter or a model is changed, and data cannot be distributed or analyzed at the same time.

In addition, the system using the 5why analysis tool of the prior art mainly collects the knowledge of experts and relies on the input model, there is a problem in limiting in-depth analysis of the cause of the error and can not solve the potential error.

Korean Laid-Open Patent No. 2017-0035791

An object of the present invention, while performing the distributed learning for classification, analysis, prediction and diagnosis of big data collected from plant control / monitoring / production facilities, checklist for 4M (Man / Machine / Material / Method) analysis And by performing the learning for 5-way analysis, to provide an analysis of the error detection, error cause, error generation line, the occurrence value and prediction value for the staff information and process information.

One embodiment of the present invention for achieving the technical problem is a statistical quality management system using big data analysis, the collection unit for indexing the analysis target data, learning data and learning model from the collection equipment to load into the storage; An analysis unit which analyzes whether the analysis target data and the training target data are suitable for a learning model and generates analysis information; A learning unit performing deep learning learning by matching the analysis target data and the learning target data with the learning model according to the analysis information; Maintenance history management unit for quantifying the maintenance history for each of the analysis target data and the learning target data corresponding to the equipment to be collected based on the deep learning learning results; And a detection complementary unit for measuring reliability and precision by analyzing correlations between causality, precision, errors, and defects about the equipment to be collected and the maintenance resume.

Preferably, the collection unit indexes a suitable learning model by indexing the data format of the analysis target data and the learning target data, and the data format may include any one of an Excel file, a csv file, or a txt file.

The learning model is characterized in that it is composed of any one of K-means, cluster or K-neighbor.

The learning unit may be configured to distribute and perform deep learning learning on the analysis target data and the learning target data to correspond to each learning model.

The learning unit is characterized by updating the maintenance resume by learning the maintenance resume through 4M (Man / Machine / Material / Method) analysis and 5 way analysis.

Statistical quality control method using big data analysis according to an embodiment of the present invention based on the system described above, the collector indexes the analysis data, the learning data and the learning model from the collection target equipment (load) to the storage ( a) step; (B) generating an analysis information by analyzing whether the analysis object data and the learning object data are suitable for a learning model; And (c) performing, by the learning unit, deep learning learning by matching the analysis target data and the learning target data with the learning model according to the analysis information.

Preferably, the step (a) includes the step of collecting and analyzing the data to be analyzed and the data to be analyzed and collected from the equipment to be collected by the collector (a-1); (A-2) the collecting unit classifying the data format of the analysis target data and the learning target data; And (a-3) indexing a learning model suitable for the data format classified by the collector.

(c) step (c-1) in which the learning unit indexes the learning definition data; (C-2) the learning unit matching each of the analysis target data, the study target data, and the learning model based on the analysis information; And (c-3) the learning unit performing deep learning learning on the analysis target data and the learning target data for each matched learning model.

(d) after the step (c), the maintenance history management unit generates a maintenance record that quantifies the maintenance history for each of the analysis target data and the learning target data corresponding to the collection target equipment based on the deep learning learning result; And (e) measuring the reliability and precision by analyzing the correlation between the causality, precision, error, and defects of the collecting equipment and the maintenance resume.

According to the present invention as described above, while performing the distributed learning for classification, analysis, prediction and diagnosis of big data collected from the plant control / monitoring / production facilities, for 4M (Man / Machine / Material / Method) analysis By performing the learning for the checklist and the 5-way analysis, there is an effect of providing an analysis of occurrence and prediction values for error detection, error cause, error occurrence line, staff information and process information.

1 is a block diagram showing a statistical quality management system using big data analysis according to an embodiment of the present invention.
2 is a diagram illustrating a configuration of an interface unit of a statistical quality management system using big data analysis according to an embodiment of the present invention.
3 is a diagram illustrating an operation flow of a statistical quality control system using big data analysis according to an embodiment of the present invention.
4 is an exemplary diagram illustrating a concept of a statistical quality management system using big data analysis according to an embodiment of the present invention.
5 is a block diagram illustrating a framework of a statistical quality management system using big data analysis according to an embodiment of the present invention.
6 is a flow chart illustrating a statistical quality management method using big data analysis according to an embodiment of the present invention.
7 is a flowchart illustrating step S100 of a method for statistical quality control using big data analysis according to an embodiment of the present invention.
8 is a flowchart illustrating step S300 of a method for statistical quality management using big data analysis according to an embodiment of the present invention.
9 is a flowchart illustrating a process after step S100 of a method for statistical quality control using big data analysis according to an embodiment of the present invention.

Specific features and advantages of the present invention will become more apparent from the following detailed description based on the accompanying drawings. Prior to this, terms or words used in the present specification and claims are defined in the technical spirit of the present invention on the basis of the principle that the inventor can appropriately define the concept of the term in order to explain the invention in the best way. It should be interpreted to mean meanings and concepts. In addition, when it is determined that the detailed description of the known function and its configuration related to the present invention may unnecessarily obscure the subject matter of the present invention, it should be noted that the detailed description is omitted.

1 is a block diagram showing a statistical quality management system (S) using big data analysis according to an embodiment of the present invention.

As shown in Figure 1, the statistical quality management system (S) using big data analysis according to an embodiment of the present invention, the collection unit 100, analysis unit 200, learning unit 300, maintenance history It is configured to include a management unit 400 and the detection supplement 500.

First, the collecting unit 100 indexes the analysis target data and the learning target data from the collection target equipment, loads them into the storage, and indexes a learning model suitable for the indexed data format.

In this case, the indexed data format is preferably understood as data generated in any one of an Excel file, a csv file, or a txt file, but one embodiment of the present invention is not limited thereto, and the jpg file, jpeg file, or pdf file is not limited thereto. It can be generated in any one of the formats.

In addition, the storage may be configured as either a memory DB or a memory provided directly in the server without a separate database.

In addition, the learning model is classified to be suitable for each data type of Excel file, csv file, txt file, jpg file, jpeg file or pdf file, but the learning model is composed of any one of K-means, cluster or K-neighbor. Can be.

The analysis unit 200 generates analysis information by analyzing whether the analysis target data and the learning target data received from the collection unit are suitable for any learning model.

In this case, the analysis information includes histograms, plots, binomial distributions, poisson distributions, probability density functions, and exponential functions for the analysis data and the learning data. It is preferable to include a numerical value for any one of an exponential function, a control chart, a sampling inspection, or a sensory test.

The learning unit 300 matches the analysis target data and the learning target data with the learning model according to the analysis information, and performs deep learning learning for each matched learning model.

That is, the learning unit 300 according to an embodiment of the present invention selects the learning model optimized for the analysis target data and the data format of the learning target data and distributes the learning models to each learning model to perform learning.

In this case, the learning unit 300 is configured to perform learning according to the learning definition data including parameters, optimization logic, columns of data to be learned, sampling methods, data to be distributed, and storage paths of distributed processing results. For example, different sampling of learning definition data can be used for learning.

In addition, the learner 300 may change the setting of the learning definition data in the json format, and thus may perform deep learning learning without reconfiguring or re-developing the program logic.

In addition, the learning unit 300 learns a maintenance resume approved by the maintenance history management unit 400 and updates the maintenance resume updated through the 4M (Man / Machine / Material / Method) analysis and 5 way analysis to the maintenance history management unit 400. Is authorized.

At this time, the learning unit 300 may perform an index calculation, cause tracking or probability test for the analysis target data and the learning target data corresponding to the collection target equipment through the maintenance resume learning.

The maintenance history management unit 400 generates, stores, and manages a maintenance history that quantifies the maintenance history for each of the analysis target data and the learning target data corresponding to the collection target equipment based on the deep learning learning result of the learning unit 300.

In this case, the maintenance resume is composed of numerical values to enable deep learning. Therefore, in the case of accumulating knowledge information by human intervention or requiring qualitative evaluation, it is possible to quantify and predict and diagnose, and can provide various information for decision making in numerical value to help the manager's judgment quickly.

The detection compensator 500 measures reliability and precision by analyzing correlations between causality, precision, errors, and defects about the equipment to be collected and maintenance records based on the values included in the learning definition data.

In this case, the reliability and precision measurement is derived through a 5-way analysis checklist for 4M (Man / Machine / Material / Method) analysis for each of the analysis target data and the learning target data corresponding to the equipment to be collected.

The detection compensator 500 may provide a prediction result for potential error of the analysis target data and the learning target data corresponding to the collection target device based on the value included in the deep learning learning result of the learner 300.

According to the statistical quality management system (S) using big data analysis according to an embodiment of the present invention as described above, it is performed by distributing learning for classification, analysis, prediction and diagnosis of data collected from the collection target equipment. By performing checklist for 4M (Man / Machine / Material / Method) analysis and learning for 5way analysis, error detection, error cause, error occurrence line, staff information and process information for the equipment to be collected Numerical and predictive values can be analyzed.

In addition, the statistical quality management system S using big data analysis according to an embodiment of the present invention may further include an interface unit 600 as shown in FIG. 2.

The interface unit 600 is a collection unit 100, the analysis unit 200, the learning unit 300, maintenance history management unit 400 and detection complementary unit 500 performing the results of each configuration of the image or video chart and It is configured to output the analysis data.

In this case, the interface unit 600 is configured to output charts and analysis data in the form of images or videos through any one module among scatter, regression, cluster, heatmap, PCA, factor analysis, kmean or kneigbor.

In addition, the interface unit 600 may transmit charts and analysis data in an image or video form to a web, an app, a PC, or a mobile device connected through an information communication network through a single UI.

3 is a diagram illustrating an operation flow of a statistical quality management system S using big data analysis according to an embodiment of the present invention. Referring to Figure 3 to examine the operation flow as follows.

1) Various model information, parameters, optimization methods, distributed server information, sampling information, and the like for deep learning are received (step 1).

2) Load training data on existing errors to detect potential errors (step 2).

3) The classification engine and the analysis engine process the classification and analysis and process and store the data in the memory DB engine (step 3). At this time, 4m analysis and 5why analysis can be performed as needed.

4) Using the learned information and the information through 4m and 5why analysis, it provides various predicted information with reliability and precision (step 4).

5) Analyze the cause of the defect or error (step 5).

6) Track expected potential errors to provide reliability and precision (step 6).

7) Visualize and output the reliability and precision (step 7).

Hereinafter, an exemplary diagram illustrating a concept of a statistical quality management system S using big data analysis according to an embodiment of the present invention will be described with reference to FIG. 4. FIG. 5 is big data analysis according to an embodiment of the present invention. It is a block diagram showing the framework of the statistical quality management system (S) using.

As shown in Figures 4 and 5, one embodiment of the present invention loads data for preferences, training sets and supervised training sets from the repository, and performs learning through distributed processing.

At this time, the learning is performed by indexing the appropriate learning model, and the learning results are stored in the memory-based DB.

And, it is configured to perform the analysis, 4m and 5why and output the result through the visualization tool.

Hereinafter, a statistical quality management method using big data analysis according to an embodiment of the present invention will be described with reference to FIG. 6.

First, the collection unit 100 indexes the analysis target data, the learning target data, and the learning model from the collection target equipment and loads them in the storage (S100).

Subsequently, the analysis unit 200 generates analysis information by analyzing whether the analysis target data and the learning target data received from the collection unit are suitable for a learning model (S200).

In addition, the learning unit 300 performs deep learning learning by matching the analysis target data and the learning target data with the learning model according to the analysis information (S300).

Hereinafter, referring to FIG. 7, the S100 step of the statistical quality control method using big data analysis according to an embodiment of the present invention is as follows.

First, the collection unit 100 indexes the analysis target data and the learning target data from the collection target equipment and loads them in the storage (S102).

Next, the collection unit 100 classifies the data format of the analysis target data and the learning target data (S104).

Then, the learning model suitable for the data format classified by the collecting unit 100 is indexed (S106).

Hereinafter, referring to FIG. 8, a detailed process of step S300 of the statistical quality control method using big data analysis according to an embodiment of the present invention will be described below.

After the step S200, the learning unit 300 indexes the learning definition data (S302).

Next, the learning unit 300 matches each of the analysis target data, the training target data, and the learning model based on the analysis information (S304).

In addition, the learning unit 300 performs deep learning learning on the analysis target data and the learning target data for each matched learning model (S306).

Hereinafter, referring to FIG. 9, the process after step S300 of the statistical quality control method using big data analysis according to an embodiment of the present invention will be described below.

After the step S300, the maintenance history management unit 400 generates a maintenance history by quantifying the maintenance history for each of the analysis target data and the learning target data corresponding to the collection target equipment based on the deep learning learning results (S400).

In addition, the detecting and compensating unit 500 analyzes the relationship between the causal relationship, the precision, the error or the defect of the equipment to be collected and the maintenance resume, and measures the reliability and the precision (S500).

As described above and described with reference to a preferred embodiment for illustrating the technical idea of the present invention, the present invention is not limited to the configuration and operation as shown and described as described above, it is a deviation from the scope of the technical idea It will be understood by those skilled in the art that many modifications and variations can be made to the invention without departing from the scope of the invention. Accordingly, all such suitable changes, modifications, and equivalents should be considered to be within the scope of the present invention.

S: Statistical Quality Management System Using Big Data Analysis
100: collector
200: analysis unit
300: learning unit
400: maintenance history management
500: detection supplement
600: interface unit

Claims (9)

A collector configured to index the analysis target data, the training target data, and the training model from the target equipment to be loaded into the storage;
An analyzer configured to analyze whether the analysis target data and the training target data are suitable for a learning model and generate analysis information;
A learning unit performing deep learning learning by matching the analysis target data and the training target data with a learning model according to the analysis information;
A maintenance history management unit for quantifying maintenance history for each of the analysis target data and the learning target data corresponding to the collection target equipment based on the deep learning learning result; And
Detecting and complementary unit for measuring the reliability and precision by analyzing the relationship between the causal relationship, precision, error or failure of the equipment and maintenance resume
Statistical quality management system using big data analysis, characterized in that it comprises. The method of claim 1,
The collection unit,
Index the appropriate learning model by indexing the data format of the analysis target data and the learning target data,
Statistical data management system using a big data analysis, characterized in that the data format includes any one of an Excel file, csv file or txt file. The method of claim 1,
The learning model,
Statistical quality management system using big data analysis, characterized in that consisting of any one of K-means, cluster or K-neighbor. The method of claim 1,
The learning unit,
Statistical quality management system using big data analysis, characterized in that for performing the deep learning learning for the analysis target data and the learning target data distributed to correspond to each of the learning model. The method of claim 1,
The learning unit,
Statistical quality management system using the big data analysis, characterized in that to update the maintenance resume through 4M (Man / Machine / Material / Method) analysis and 5-way analysis by learning the maintenance resume. (a) collecting, by the collector, the analysis target data, the training target data, and the training model from the collection target equipment and loading the data into the storage;
(b) an analysis unit generating analysis information by analyzing whether the analysis target data and the learning target data are suitable for a learning model; And
(c) a learning unit performing deep learning learning by matching the analysis target data and the learning target data with a learning model according to analysis information;
Statistical quality control method using big data analysis, comprising. The method of claim 6,
In step (a),
(a-1) a step in which the collecting unit indexes the analysis target data and the study target data from the collection target equipment and loads the collected data into the storage;
(a-2) collecting, by the collector, classifying data types of the analysis target data and the learning target data; And
(a-3) indexing the learning model for the data format classified by the collector;
Statistical quality control method using big data analysis, comprising. The method of claim 6,
In step (c),
(c-1) the learning unit indexing the learning definition data;
(c-2) the learning unit matching each of the analysis target data, the study target data, and the learning model based on the analysis information; And
(c-3) the learning unit performing deep learning learning on the analysis target data and the learning target data for each matched learning model.
Statistical quality control method using big data analysis, characterized in that it comprises. The method of claim 6,
After step (c),
(d) generating, by the maintenance history management unit, a maintenance history that quantifies the maintenance history for each of the analysis target data and the learning target data corresponding to the collection target equipment based on the deep learning learning result; And
(e) Detecting and measuring the reliability and precision by analyzing the correlation between the causality, precision, error, and defects of the equipment and maintenance records to be collected.
Statistical quality control method using big data analysis, comprising.

Images