Improving Manufacturing Plants Through Big Data Analytics SDS2019 - - PowerPoint PPT Presentation

▶
improving manufacturing plants
SMART_READER_LITE
LIVE PREVIEW

Improving Manufacturing Plants Through Big Data Analytics SDS2019 - - PowerPoint PPT Presentation

Nov 22, 2023 232 likes •538 views

Improving Manufacturing Plants Through Big Data Analytics SDS2019 14 June 2019 Prof. Dr. Kurt Stockinger Martin Weber Marc Schni Agenda Use Case Goals Architecture Blueprint Experiment Conclusions Evaluation

slide-1
SLIDE 1

Improving Manufacturing Plants Through Big Data Analytics

SDS2019 14 June 2019

  • Prof. Dr. Kurt Stockinger

Martin Weber Marc Schöni

slide-2
SLIDE 2

Agenda

  • Use Case
  • Goals
  • Architecture Blueprint
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-3
SLIDE 3

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-4
SLIDE 4

About Midor

  • Founded in 1928
  • Located in Meilen ZH
  • 600 Employees
  • Part of M-Industry
  • Produce 250’000 items daily for Migros und others
  • 32 production lines, 940 different products (different biscuits,

ice cream, snacks, dessert powder, etc.)

slide-5
SLIDE 5

Introduction

  • Production line 16 produces the Blévita, one of Midors sales hits
  • Short production stops are reducing the output
  • What causes these disturbances?
  • Can they be predicted?
slide-6
SLIDE 6

Line 16

Batter mixing Oven Slots 1-9 Packaging & Labelling

slide-7
SLIDE 7

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-8
SLIDE 8

Goals

Qualitative Goals

  • 1. Improve efficiency
  • 2. Flexible and scalable

architecture

  • 3. Allows processing of various

data formats Quantitative Goals

  • 1. Find the most relevant

features causing the disturbances

  • 2. Latency for Inference of

< 5 seconds

  • 3. Cost should be at worst

proportional to amount of processed data

slide-9
SLIDE 9

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-10
SLIDE 10

System Context

slide-11
SLIDE 11

Conceptual Architecture

slide-12
SLIDE 12

Solution Architecture

slide-13
SLIDE 13

Solution Architecture

slide-14
SLIDE 14

Solution Architecture

slide-15
SLIDE 15

Solution Architecture

A X P P

M

P

M

slide-16
SLIDE 16

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-17
SLIDE 17

Objective

  • Over a period of ~1 year data about short production stoppages

was collected (→ Label)

  • Over the same period additional data about orders,

climate conditions etc. were captured (→ Features)

  • Is it possible to find a pattern in these datasets regarding the
  • ccurrence of short stops?
slide-18
SLIDE 18

Big Data Analytics Solution

„System under Design“

FAcility Management System (FAMS)

3rd party system

Climate System

3rd party system

Manufacturing Execution System (MES)

3rd party system

Gateway Production line

3rd party system

BaroHygro.XML FAMS.XLSX Orders.XML Signal-Stream

x1, x2, … xn y

Amperes oven: 32A

  • Rel. Humidity: 54%

Product: Blévita Gruyère Short stop: Nein

Building the dataset

slide-19
SLIDE 19

Splitting the dataset

Unbalanced dataset for modelling (90%) unbalanced verification dataset (10%) Balanced dataset unbalanced dataset (100%) 10x Cross-Validation

Training Model

Train Test no stops stops

slide-20
SLIDE 20

Modelquality

Accuracy Precision Recall F1 Score AUC Random Forest .852 .818 .897 .856 .911 Gradient Boosted Tree .847 .814 .892 .852 .909 Logistic Regression .629 .616 .644 .630 .677 Support Vector Machine .612 .692 .588 .636 .659 Naïve Bayes .572 .680 .552 .610 .602

slide-21
SLIDE 21

Feature Reduction

Model quality metrics (see legend) Comparison of model quality metrics depending on number of features

number of features

slide-22
SLIDE 22

Important Features

slide-23
SLIDE 23

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-24
SLIDE 24

Findings: Machine Learning

  • We can predict shorts stops with a F1-Score of about 85%
  • The integration of different data sources took the most time
  • It is not possible to a priori estimate the importance of

predictors/features per data source → Integrate all data sources

  • The prediction itself does not provide a business value without

additional steps (operationalization)

slide-25
SLIDE 25

Conclusion: Data import

Data source 1 Target Dataset (combination) Data source 2 Data source 3 Events

? ? ?

Time

slide-26
SLIDE 26

Conclusions: Architecture

  • Findings from modelling provide new boundary conditions for the

big data architecture:

  • Number and kind of data sources
  • Amount of data
  • Requirements for inference service (Compute, Memory)
  • Principles of Lambda-Architecture have proven their effectiveness
  • Benefits of Kapa architecture (single code base) using libraries
  • Tools for ML-Pipeline export & operationalization are in early stages
  • Monitoring of data quality is a crucial success factor
slide-27
SLIDE 27

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion
slide-28
SLIDE 28

Evaluation

Quality Goals

  • 1. Improve efficiency
  • 2. Flexible and scalable

architecture

  • 3. Allows processing of various

data formats Quantity Goals

  • 1. Find the most relevant

Features causing the disturbances

  • 2. Latency for Inference of

< 5 seconds

  • 3. Cost should be at worst

proportional to amount of processed data

*

slide-29
SLIDE 29

Agenda

  • Use Case
  • Goals
  • Solution Architecture
  • Experiment
  • Conclusions
  • Evaluation
  • Q&A / Discussion