Taguchi. Abstraction Optimisation of manufacturing processes is typically performed utilising mathematical process models or designed experiments. However,

Slides:



Advertisements
Similar presentations
Principal Component Analysis Based on L1-Norm Maximization Nojun Kwak IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008.
Advertisements

1 An Adaptive GA for Multi Objective Flexible Manufacturing Systems A. Younes, H. Ghenniwa, S. Areibi uoguelph.ca.
Application of Heuristics and Meta-Heuristics Scheduling: Job Shop Scheduling Parallel Dedicated Machine (PDS1) Single machine minimize total tardiness.
CSCI 347 / CS 4206: Data Mining Module 07: Implementations Topic 03: Linear Models.
CHAPTER 21 Inferential Statistical Analysis. Understanding probability The idea of probability is central to inferential statistics. It means the chance.
Mathematical Analysis of Robustness Sensitivity analysis allows the linking of robustness to network structure. However, it yields only local properties.
Experimental Design, Response Surface Analysis, and Optimization
Gizem ALAGÖZ. Simulation optimization has received considerable attention from both simulation researchers and practitioners. Both continuous and discrete.
1 Multivariate Statistics ESM 206, 5/17/05. 2 WHAT IS MULTIVARIATE STATISTICS? A collection of techniques to help us understand patterns in and make predictions.
Date:2011/06/08 吳昕澧 BOA: The Bayesian Optimization Algorithm.
Factor Analysis Purpose of Factor Analysis
Genetic algorithms for neural networks An introduction.
Reporter : Mac Date : Multi-Start Method Rafael Marti.
A Comparative Study Of Deterministic And Stochastic Optimization Methods For Integrated Design Of Processes Mario Francisco a, Silvana Revollar b, Pastora.
Supply Chain Design Problem Tuukka Puranen Postgraduate Seminar in Information Technology Wednesday, March 26, 2009.
Bayesian belief networks 2. PCA and ICA
Nonlinear Stochastic Programming by the Monte-Carlo method Lecture 4 Leonidas Sakalauskas Institute of Mathematics and Informatics Vilnius, Lithuania EURO.
Uncertainty in Wind Energy
Microarray Gene Expression Data Analysis A.Venkatesh CBBL Functional Genomics Chapter: 07.
1 Statistical Tools for Multivariate Six Sigma Dr. Neil W. Polhemus CTO & Director of Development StatPoint, Inc. Revised talk:
Data Mining Techniques
1 Reasons for parallelization Can we make GA faster? One of the most promising choices is to use parallel implementations. The reasons for parallelization.
U N I V E R S I T À D E G L I S T U D I D I M I L A N O C17 SC for Environmental Applications and Remote Sensing I M S C I A Soft Computing for Environmental.
Review of Statistical Inference Prepared by Vera Tabakova, East Carolina University ECON 4550 Econometrics Memorial University of Newfoundland.
The Tutorial of Principal Component Analysis, Hierarchical Clustering, and Multidimensional Scaling Wenshan Wang.
By Paul Cottrell, BSc, MBA, ABD. Author Complexity Science, Behavioral Finance, Dynamic Hedging, Financial Statistics, Chaos Theory Proprietary Trader.
A Genetic Algorithms Approach to Feature Subset Selection Problem by Hasan Doğu TAŞKIRAN CS 550 – Machine Learning Workshop Department of Computer Engineering.
1 Validation & Verification Chapter VALIDATION & VERIFICATION Very Difficult Very Important Conceptually distinct, but performed simultaneously.
by B. Zadrozny and C. Elkan
An Approach of Artificial Intelligence Application for Laboratory Tests Evaluation Ş.l.univ.dr.ing. Corina SĂVULESCU University of Piteşti.
Slides are based on Negnevitsky, Pearson Education, Lecture 12 Hybrid intelligent systems: Evolutionary neural networks and fuzzy evolutionary systems.
A two-stage approach for multi- objective decision making with applications to system reliability optimization Zhaojun Li, Haitao Liao, David W. Coit Reliability.
Part 3 Managing for Quality and Competitiveness © 2015 McGraw-Hill Education.
Stochastic Linear Programming by Series of Monte-Carlo Estimators Leonidas SAKALAUSKAS Institute of Mathematics&Informatics Vilnius, Lithuania
1 MULTIVARIATE OPTIMIZATION CONSIDERING QUALITY AND MANUFACTURING COSTS: A CASE STUDY IN A DRYING PROCESS Carla Schwengber ten Caten PPGEP/UFRGS – BRAZIL.
(Particle Swarm Optimisation)
Repeated Measurements Analysis. Repeated Measures Analysis of Variance Situations in which biologists would make repeated measurements on same individual.
GENETIC ALGORITHMS FOR THE UNSUPERVISED CLASSIFICATION OF SATELLITE IMAGES Ankush Khandelwal( ) Vaibhav Kedia( )
Genetic Algorithms Introduction Advanced. Simple Genetic Algorithms: Introduction What is it? In a Nutshell References The Pseudo Code Illustrations Applications.
Digital Media Lab 1 Data Mining Applied To Fault Detection Shinho Jeong Jaewon Shim Hyunsoo Lee {cinooco, poohut,
CHAPTER 12 Descriptive, Program Evaluation, and Advanced Methods.
Descriptive Statistics vs. Factor Analysis Descriptive statistics will inform on the prevalence of a phenomenon, among a given population, captured by.
28/09/2011Ivan S. Živković1 Artificial Neural Networks for Decision Support in Copper Smelting Process Ivan S. Živković Mathematical Institute of the Serbian.
2005/12/021 Content-Based Image Retrieval Using Grey Relational Analysis Dept. of Computer Engineering Tatung University Presenter: Tienwei Tsai ( 蔡殿偉.
Project 11: Determining the Intrinsic Dimensionality of a Distribution Okke Formsma, Nicolas Roussis and Per Løwenborg.
Project 11: Determining the Intrinsic Dimensionality of a Distribution Okke Formsma, Nicolas Roussis and Per Løwenborg.
Tetris Agent Optimization Using Harmony Search Algorithm
Genetic Algorithms Abhishek Sharma Piyush Gupta Department of Instrumentation & Control.
Reservoir Uncertainty Assessment Using Machine Learning Techniques Authors: Jincong He Department of Energy Resources Engineering AbstractIntroduction.
Module III Multivariate Analysis Techniques- Framework, Factor Analysis, Cluster Analysis and Conjoint Analysis Research Report.
Chapter 20 Classification and Estimation Classification – Feature selection Good feature have four characteristics: –Discrimination. Features.
Alice E. Smith and Mehmet Gulsen Department of Industrial Engineering
Data Mining and Decision Support
D Nagesh Kumar, IIScOptimization Methods: M8L5 1 Advanced Topics in Optimization Evolutionary Algorithms for Optimization and Search.
A field of study that encompasses computational techniques for performing tasks that require intelligence when performed by humans. Simulation of human.
1 Tom Edgar’s Contribution to Model Reduction as an introduction to Global Sensitivity Analysis Procedure Accounting for Effect of Available Experimental.
Agenda  INTRODUCTION  GENETIC ALGORITHMS  GENETIC ALGORITHMS FOR EXPLORING QUERY SPACE  SYSTEM ARCHITECTURE  THE EFFECT OF DIFFERENT MUTATION RATES.
Artificial Intelligence By Mr. Ejaz CIIT Sahiwal Evolutionary Computation.
TAUCHI PHILOSOPHY SUBMITTED BY: RAKESH KUMAR ME
CPH Dr. Charnigo Chap. 11 Notes Figure 11.2 provides a diagram which shows, at a glance, what a neural network does. Inputs X 1, X 2,.., X P are.
An Evolutionary Algorithm for Neural Network Learning using Direct Encoding Paul Batchis Department of Computer Science Rutgers University.
Breeding Swarms: A GA/PSO Hybrid 簡明昌 Author and Source Author: Matthew Settles and Terence Soule Source: GECCO 2005, p How to get: (\\nclab.csie.nctu.edu.tw\Repository\Journals-
A PID Neural Network Controller
Methods of multivariate analysis Ing. Jozef Palkovič, PhD.
 Negnevitsky, Pearson Education, Lecture 12 Hybrid intelligent systems: Evolutionary neural networks and fuzzy evolutionary systems n Introduction.
Evolutionary Computation Evolving Neural Network Topologies.
01-Business intelligence
Balancing of Parallel Two-Sided Assembly Lines via a GA based Approach
Descriptive Statistics vs. Factor Analysis
Somi Jacob and Christian Bach
Presentation transcript:

Taguchi

Abstraction Optimisation of manufacturing processes is typically performed utilising mathematical process models or designed experiments. However, such approaches could not be used in case when explicit quality function is unknown and when actual experimentation would be expensive and time-consuming. Taguchi

Abstraction (continued) The paper presents an approach to optimisation of manufacturing processes with multiple potentially correlated responses, using historical process data. The integrated approach is consisted from Two methods: the first relays on Taguchi’s quality loss function and multivariate statistical methods, the,,, Taguchi

Abstraction (continued) The second method is based on the first one and employs artificial neural networks and genetic algorithm to ensure global optimal settings of a critical parameters found in a continual space of solutions. The case study of a multi-response process with correlated responses was used to illustrate the effective application of the proposed approach, where historical data collected during normal production and stored in a control charts were used for process optimisation. Taguchi

Introduction Process optimisation is typically performed by analysing the process responses obtained from designed experiments, carried out on the actual manufacturing process. But, conducting experiments on the actual process tends to cause distruption in the plant and may be uneconomic. The possibility to use process historical data (i.e. from the control charts) has not been explored videly in the literature. There are few studies that used historical data for optimisation, but they discuss only singleresponse problems. Taguchi

Introduction (Continued) Several characteristics of a product are usually considered for product quality by the customer. In such cases, a single optimum setting of process parameters needs to be identified so that the specifications of all quality haracteristics (responses) are met. Complexity of the problem increases when the responses are correlated. Taguchi

The factor effects method Taguchi’s quality loss function financial measure of the customer dissatisfaction with a product's performance as it deviates from a target value. Unlike the conventional weighting methods, the quality loss function adequately presents relative financial significance of responses, thus providing a right metric for multicriteria decision making. Taguchi

The factor effects method, Taguchi’s quality loss function continued… Taguchi The quality loss of the i-th quality characteristic in the k-th point QLik

The factor effects method, Principal component analysis (PCA) PCA is considered as an effective means of transforming correlated responses into uncorrelated linear combinations (principal components). In the presented approach, PCA is performed on NQL data resulting in a set of uncorrelated components. Taguchi

The factor effects method, Grey Relational Analysis (GRA) continued… GRA provides an effective means of dealing with one event that involves multiple decisions and deals with poor, incomplete and uncertain data. In the presented approach, GRA is performed on the absolute value of principal component scores Yi(k). Linear preprocessing method is employed to transform the principal component scores |Yi(k)| into a set of standardised multi-response performance statistics Zi(k): Taguchi

Grey Relational Analysis (GRA) continued Taguchi

The ANN&GA-based method Taguchi ANN is powerful technique to generate complex multi-response, linear and non-linear process models without referring to a particular mathematical model, proven as effective in various applications. GA was chosen for optimisation because it is proven as a potent multiple-directional heuristic search method for optimising highly nonlinear.

Implementation of the factor effects method Taguchi

Implementation of the factor effects method Taguchi

Topology of Anns Taguchi

Result of Anns Taguchi

CONCLUSION The paper presented two methods for multi- response process optimisation for correlated responses, which employ historical data 1. By using Taguchi’s SN ratio and quality loss, relative significances of responses are adequately represented and the response mean and variation are assessed simultaneously. 2. Multivariate statistical methods PCA and GRA are employed to uncorrelate and synthesise responses, ensuring that the Taguchi

CONCLUSION (cantinued) 1. The GA’s capacity of performing global search among all solutions in continual multi dimensional space ensures convergence to the global optimal parameter settings. 2. The initial population in GA is formed in the proximity of the potentially good solution (the parameter settings obtained by the factor effects method), which advances the convergence to the global solution, meaning that the probability of finding the actual global parameters solution in the given number of generation Taguchi

CONCLUSION (cantinued),,,,is significantly improved. If the initial population was not defined at such way (e.g. if the initial population was randomly generated), in general, GA might not be able to find the actual global solution in a limited number of iterations. 3. The proposed method does not depend on the type of the relations between responses and critical parameters, type and number of process parameters and responses, existence of correlations between responses or process parameters, or their interrelations. Taguchi

Six Sigma