WO2016103611A1 - Dispositif d'analyse de facteurs, procédé d'analyse de facteurs et programme d'analyse de facteurs - Google Patents

Dispositif d'analyse de facteurs, procédé d'analyse de facteurs et programme d'analyse de facteurs Download PDF

Info

Publication number
WO2016103611A1
WO2016103611A1 PCT/JP2015/006179 JP2015006179W WO2016103611A1 WO 2016103611 A1 WO2016103611 A1 WO 2016103611A1 JP 2015006179 W JP2015006179 W JP 2015006179W WO 2016103611 A1 WO2016103611 A1 WO 2016103611A1
Authority
WO
WIPO (PCT)
Prior art keywords
series data
time series
time
predetermined number
factor analysis
Prior art date
Application number
PCT/JP2015/006179
Other languages
English (en)
Japanese (ja)
Inventor
広晃 福西
毅彦 溝口
Original Assignee
日本電気株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電気株式会社 filed Critical 日本電気株式会社
Priority to JP2016565890A priority Critical patent/JP6841039B2/ja
Publication of WO2016103611A1 publication Critical patent/WO2016103611A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the present invention relates to a factor analysis device, a factor analysis method, and a program recording medium, in particular, a factor analysis device, a factor analysis method, and a factor analysis method for estimating a factor involved in a difference in characteristics such as quality of a manufactured product from time series data.
  • the present invention relates to a program recording medium.
  • Japanese Patent Application Laid-Open No. 2004-228688 creates a relational model that represents the relationship between the amount of work and factors that affect the amount of work for each business group, and evaluates the stability of the work amount of the business group based on the model. Is disclosed.
  • the apparatus creates a relational model by Markov transition multivariate autoregressive analysis.
  • a wide variety of sensors are installed in production plants and equipment, and time-series data is recorded.
  • the amount of data of the time series data per sensor increases as the acquisition period is longer. Further, the total amount of data obtained from all sensors increases as the number of sensors increases.
  • the present invention aims to provide a factor analysis device, a factor analysis method, and a program for reducing the above-described problems.
  • the factor analysis apparatus is obtained from two or more sources that are divided into a temporal range (hereinafter referred to as a region) given as one of two or more characteristics as an attribute.
  • Time series data storage means for storing original time series data as time series data of measured values, and extracting one or more sections obtained by dividing the area into smaller time ranges from each area, and time progress Reduced data generating means for generating a predetermined number of shortened time series data by combining in order, and factor estimating means for estimating the source involved in the difference in characteristics from each shortened time series data and outputting as an individual estimation result
  • integrated determination means for obtaining an integrated estimation result obtained by integrating the predetermined number of individual estimation results.
  • the machine-readable recording medium includes two or more divided into a time range (hereinafter referred to as an area) provided with any one of the two or more characteristics as an attribute.
  • a program for causing a computer to execute a process for obtaining an integrated estimation result obtained by integrating the individual estimation results is stored.
  • the factor analysis apparatus can suppress an increase in the number of main memories and a decrease in analysis accuracy that lead to an increase in cost in factor analysis based on a large amount of time-series data.
  • the original time series data is divided into regions for each temporal range from which data was acquired. That is, the original time series data is divided such that data measured from time t to time t + ⁇ 1 is region 1, data measured from time t + ⁇ 1 to time t + ⁇ 1 + ⁇ 2 is region 2, and so on. ing. Note that the time length of each area may be constant or may not be constant.
  • the factor analysis apparatus 20 When the original time series data is input, the factor analysis apparatus 20 1) generates a predetermined number K sets of shortened time series data, 2) performs factor estimation for each shortened time series data, and 3) each shortened time series. Integration of estimation results obtained from series data.
  • the factor analysis apparatus 20 generates a predetermined number K (K is a plurality) of different shortened time series data. This predetermined number K is given to the factor analysis device 20 by an administrator or the like as an input parameter.
  • the factor analysis device 20 may output only the sensor that is estimated to be most related to the quality of the product as shown in Table 2.
  • Table 2 shows the sensors estimated in the “estimated m” column from the m-th shortened time series data that are most likely to be related to the quality of the product.
  • the factor analysis device 20 integrates a predetermined number K of estimation results.
  • the factor analysis device 20 obtains the ratio of the acquisition rank of each sensor as shown in Table 3 with respect to the output shown in Table 1, and extracts the sensor with the highest ratio in each rank to determine the integration rank. Can do.
  • the integration order is as follows.
  • the first place is the sensor 3 with a ratio of 30%.
  • the second place is the sensor 1 with a ratio of 15%.
  • the third place is the sensor 2 with a ratio of 12%.
  • the factor analysis device 20 may determine the subsequent ranks in the same manner.
  • the factor analysis device 20 may determine the integration order for the outputs shown in Tables 1 and 3 by other methods. For example, the factor analysis device 20 sets a score for each rank, obtains an integrated score by summing up the product of the score and the rate at which each sensor in Table 3 has acquired each rank to a predetermined rank, The integration rank of each sensor may be determined in descending order.
  • FIG. 4 is a configuration diagram of the factor analysis device 20 according to the first embodiment.
  • the factor analysis device 20 is connected to the input device 10 and the output device 40.
  • the factor analysis device 20 includes a shortened data generation unit 21, a factor estimation unit 22, an integration determination unit 23, a time series data storage unit 31, and a shortened data storage unit 32.
  • the direction of the arrow in the drawing shows an example of the flow of control / data, and does not limit the direction of the signal between blocks.
  • the time series data storage unit 31 stores original time series data to which information indicating the boundary and characteristics of the input area and the boundary of the section is added.
  • the specification of the area range of the original time series data may be specified by time or may be specified by the acquisition timing serial number (also referred to as the number of data points).
  • the designation by time is, for example, 0-1 hour (high quality), 1-2 hours (low quality).
  • the designation by the number of data points is, for example, 1-100 points (high quality) and 101-200 points (low quality).
  • the time series data per sensor is 160,000 points.
  • Information obtained from these sensors, information indicating the boundary and characteristics of the region, time series data to which information indicating the boundary of the section is added, and a predetermined number K are input from the input device 10 to the factor analysis device 20. .
  • FIG. 6 shows time-series data input to the factor analysis device 20 in this specific example.
  • FIG. 6 shows a group of measurement values of eight sensors from A to H by square marks, and the time series time advances from left to right.
  • the abbreviated data generation unit 21 of the factor analysis device 20 extracts one section from each region at random and combines them in the order of time progression.
  • 50 is designated as the predetermined number K
  • the shortened data generation unit 21 generates 50 sets of shortened time-series data of 8 sensors.
  • One set of shortened time-series data generated here is of a size that can be processed by the main memory.
  • the factor estimating unit 22 converts the shortened time series data into a feature amount, for example, a moving average time series, and performs multivariate analysis on the converted shortened time series data.
  • the factor estimating unit 22 uses, for example, logistic regression as an analysis algorithm.
  • the effect in this case is as follows.
  • the shortened time series data generated by the factor analysis device 20 is 1/10 the size of the original time series data. Since these 50 sets of shortened time series data are distributed to 8 arithmetic units, multivariate analysis of 6 or 7 sets of shortened time series data is performed per arithmetic unit. As a result, the calculation is completed in at least about 6 to 10 times as long as the time series data is calculated by one arithmetic unit.
  • the factor estimator 22 outputs a plurality of sensors with rankings that are presumed to be largely related to quality, as factor candidate estimation results.
  • Table 4 shows the percentage of each sensor that acquired the first, second, and third positions in the estimation results based on the 50 shortened time-series data calculated by the integration determination unit 23.
  • the integration determination unit 23 of the factor analyzer 20 sums up the product of the score of each rank and the ratio of the acquisition rank of each sensor up to a predetermined rank to obtain an integrated score, and determines the integration rank in descending order of the integration score. Also good. For example, when 8 points are given to the 1st place, 7 points are given to the 2nd place, and 6 points are given to the 3rd place, the integration determining unit 23 calculates the integrated score as follows.
  • the score given to the ranking may be a lower value as the ranking is higher, and the integration ranking may be lower in the order of the integration score.
  • the source of measurement values is not limited to sensors.
  • the generation source may be, for example, a human such as an investigator.
  • the generation source may be an information terminal device that outputs data such as stock prices and exchange rates.
  • the factor analysis device 20 can suppress an increase in main memory that leads to an increase in cost in factor analysis based on a large amount of time-series data.
  • the reason is that the factor analysis device 20 generates a predetermined number of shortened time series data from the original time series data, integrates the estimation results obtained from the respective shortened time series data, and obtains an integrated estimation result. is there.
  • Each shortened time series data can be analyzed with a small amount of main memory compared to the original time series data.
  • the factor analysis device 20 can output an estimation result with the same accuracy as the analysis of the original time series data. The reason is that the factor analysis device 20 integrates estimation results obtained from a predetermined number of shortened time series data to obtain an integrated estimation result.
  • FIG. 7 is a configuration diagram of the factor analysis device 20 according to the second embodiment.
  • the factor analysis device 20 according to the present embodiment includes a shortened data generation unit 21, a factor estimation unit 22, an integration determination unit 23, and a time series data storage unit 31.
  • the time-series data storage unit 31 is a measurement value obtained from two or more sources divided into a temporal range (hereinafter referred to as a region) given as one of the two or more characteristics as an attribute.
  • the original time-series data which is the time-series data, is stored.
  • the abbreviated data generation unit 21 extracts one or more sections obtained by dividing the area into smaller time ranges from each area and combines them in the order of time progress to generate a predetermined number of reduced time series data.
  • the factor estimator 22 estimates the generation source involved in the difference in characteristics from each shortened time series data and outputs it as an individual estimation result.
  • the integration determination unit 23 obtains an integrated estimation result obtained by integrating a predetermined number of individual estimation results.
  • the factor analysis device 20 can suppress an increase in main memory that leads to an increase in cost in factor analysis based on a large amount of time-series data.

Landscapes

  • Engineering & Computer Science (AREA)
  • Manufacturing & Machinery (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Automation & Control Theory (AREA)
  • Testing And Monitoring For Control Systems (AREA)
  • General Factory Administration (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

L'objectif de la présente invention est de fournir une analyse de facteurs sur la base d'une grande quantité de données de séries temporelles, avec une extension minimale de la mémoire principale, ce qui aurait conduit à une augmentation de coût, et une détérioration minimale dans la précision d'analyse. Ce dispositif d'analyse de facteurs est pourvu d'un moyen de génération de données tronquées, d'un moyen d'extrapolation de facteur, d'un moyen d'intégration/détermination, ainsi que d'un moyen de stockage de données de séries temporelles. Le moyen de stockage de données de séries temporelles stocke des données de séries temporelles d'origine qui sont destinées à des valeurs de mesure obtenues à partir d'au moins deux sources de génération et qui sont divisées en plages temporelles (ci-après désignées en tant que régions), à chacune étant attribué n'importe laquelle d'au moins deux caractéristiques en tant qu'un attribut de celle-ci. Le moyen de génération de données tronquées extrait à partir de chaque région au moins un segment généré par division de cette région en plages temporelles plus petites, et combine des segments dans un ordre temporel afin de générer un nombre prédéterminé d'ensembles de données de séries temporelles tronquées. Le moyen d'extrapolation de facteurs extrapole la source de génération qui contribue à la différence de caractéristiques dans chaque ensemble de données de séries temporelles tronquées et délivre en sortie ces sources de génération en tant que résultats d'extrapolation individuels. Le moyen d'intégration/détermination obtient un résultat d'extrapolation intégré dans lequel le nombre prédéterminé de résultats d'extrapolation individuels est intégré.
PCT/JP2015/006179 2014-12-22 2015-12-11 Dispositif d'analyse de facteurs, procédé d'analyse de facteurs et programme d'analyse de facteurs WO2016103611A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2016565890A JP6841039B2 (ja) 2014-12-22 2015-12-11 要因分析装置、要因分析方法、および、プログラム

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2014-259157 2014-12-22
JP2014259157 2014-12-22

Publications (1)

Publication Number Publication Date
WO2016103611A1 true WO2016103611A1 (fr) 2016-06-30

Family

ID=56149680

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2015/006179 WO2016103611A1 (fr) 2014-12-22 2015-12-11 Dispositif d'analyse de facteurs, procédé d'analyse de facteurs et programme d'analyse de facteurs

Country Status (2)

Country Link
JP (1) JP6841039B2 (fr)
WO (1) WO2016103611A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018096683A1 (fr) * 2016-11-28 2018-05-31 日本電気株式会社 Dispositif, procédé et programme d'analyse factorielle

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002312430A (ja) * 2001-04-16 2002-10-25 Nippon Steel Corp 操業分析装置
JP2006040181A (ja) * 2004-07-29 2006-02-09 Sharp Corp データ分析装置およびデータ分析方法並びにデータ分析プログラム
JP2013210945A (ja) * 2012-03-30 2013-10-10 Toshiba Corp 波形分析装置および波形分析方法
JP2014179060A (ja) * 2013-02-18 2014-09-25 Kobe Steel Ltd 品質異常の原因推定支援システム

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002312430A (ja) * 2001-04-16 2002-10-25 Nippon Steel Corp 操業分析装置
JP2006040181A (ja) * 2004-07-29 2006-02-09 Sharp Corp データ分析装置およびデータ分析方法並びにデータ分析プログラム
JP2013210945A (ja) * 2012-03-30 2013-10-10 Toshiba Corp 波形分析装置および波形分析方法
JP2014179060A (ja) * 2013-02-18 2014-09-25 Kobe Steel Ltd 品質異常の原因推定支援システム

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018096683A1 (fr) * 2016-11-28 2018-05-31 日本電気株式会社 Dispositif, procédé et programme d'analyse factorielle
JPWO2018096683A1 (ja) * 2016-11-28 2019-10-17 日本電気株式会社 要因分析方法、要因分析装置および要因分析プログラム

Also Published As

Publication number Publication date
JP6841039B2 (ja) 2021-03-10
JPWO2016103611A1 (ja) 2017-11-02

Similar Documents

Publication Publication Date Title
JP5871193B2 (ja) 監視データ分析装置、監視データ分析方法および監視データ分析プログラム
Suhariyanto et al. Asian agricultural productivity and convergence
JP6658250B2 (ja) 異常診断方法、異常診断装置、及び異常診断プログラム
Andrews eExports, imports, and economic growth in liberia: evidence from causality and cointegration analysis
US20190310927A1 (en) Information processing apparatus and information processing method
JP6561682B2 (ja) シリコンウェーハの工程計画立案システム、工程計画立案装置、工程計画立案方法及びプログラム
WO2016042774A1 (fr) Appareil et procédé d'estimation de distorsion et appareil et procédé de diagnostic de panne
JPWO2008111349A1 (ja) 生存分析システム、生存分析方法および生存分析用プログラム
JP2019020631A5 (ja) 楽曲解析方法、楽曲解析装置およびプログラム
WO2016103611A1 (fr) Dispositif d'analyse de facteurs, procédé d'analyse de facteurs et programme d'analyse de facteurs
KR102212875B1 (ko) 해 탐색 시스템 및 방법, 해 탐색 프로그램
JP5793228B1 (ja) 欠陥数予測装置及び欠陥数予測プログラム
US10853538B2 (en) Model generation system and model generation method
US20230214668A1 (en) Hyperparameter adjustment device, non-transitory recording medium in which hyperparameter adjustment program is recorded, and hyperparameter adjustment program
Hassanvand et al. A robust control chart for simple linear profiles in two‐stage processes
JP2018044958A (ja) バイアス推定装置、その方法およびプログラム
KR20170074418A (ko) 서열의 유사도 측정을 위한 k-mer 변환 장치 및 방법
JP2019096977A (ja) 可視化方法、可視化装置及び可視化プログラム
JPWO2020149389A1 (ja) 工程改善支援装置、工程改善支援方法および工程改善支援プログラム
Contreras et al. The Current Impact Factor and the long-term impact of scientific journals by discipline: A logistic diffusion model estimation
JP6849084B2 (ja) 情報分析装置、情報分析方法および情報分析プログラム
Marković et al. Prediction of the stock market trend using LS-SVMs based on technical indicators
JP6767356B2 (ja) モデル構築システムおよびモデル構築方法
JPWO2022185444A5 (ja) 互換性評価装置、互換性評価方法、及び、プログラム
JP2023072958A (ja) モデル生成装置、モデル生成方法及びデータ推定装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15872186

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2016565890

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15872186

Country of ref document: EP

Kind code of ref document: A1