Chapter 1 Introduction
R. J. Chang
Department of Mechanical Engineering
NCKU
§ 1.1 Problem and Analysis
§ 1.2 Data Engineering
§ 1.3 Scope
§ 1.4 Limitations of Course
§ 1.1 Problem and Analysis(1)
1.Problem Cognition
(a) Human and Physical interactions
Cognition problem
Communication
Physical World Human Being
Control
Interactions model
Q: How does the human being interacted with physical
world through data?
§ 1.1 Problem and Analysis(2)
Human cognition model
Theoretical Computer
Linguistics Science
Experimental
Psychology
§ 1.1 Problem and Analysis(3)
(b)Engineering problem
Human Society
Forward
Information Products
Reverse
Reverse problem：Engineering analysis
Forward problem：Engineering design
§ 1.1 Problem and Analysis(4)
Design problem:
1.Objective
2.Design parameters
3.Design constraints
Engineering decision making:
Cost：Expense for data collection and processing
Performance：Information extraction and service
§ 1.1 Problem and Analysis(5)
To trade off between cost and performance
Cost Performance
§ 1.1 Problem and Analysis(6)
2.Information Extraction
(a)Physical aspect：Interaction model
Physical Information
Field Instrumentation Field
Interaction issue
§ 1.1 Problem and Analysis(7)
(b)Mathematical aspect：Transformation model
Transformation
Physical Mathematical Analytical Information
(Topological
World Space Space Space
Space)
Modeling issue
§ 1.1 Problem and Analysis(8)
(c)Engineering aspect：Causality model
Physical Information
Algorithm
Data Display
Uncertainty issue
§ 1.2 Data Engineering(1)
1.Design Activity
(a)Design problem
1. Object: Max(signal/noise)
2. Controlled parameter:
(1)Hardware for data processing
(2)Software by algorithm (math + numerical)
3. Constraints:
(1) Hardware
(2) Software
(3) Data constraint:
finite data length
finite sampling rate
finite bandwidth
§ 1.2 Data Engineering(2)
(b)Stochastic signal and information
1. Amplitude: Probability density function
2. Time: Autocorrelation function
3. Frequency (Temporal, Spatial):
Power spectrum density function
§ 1.2 Data Engineering(3)
2.Data Analysis
(a)General procedure
Data Information
Signal
Algorithm Display
Source
Analysis and
Decision making
§ 1.2 Data Engineering(4)
Algorithm:
Data Data Data
Preparation Qualification Analysis
* Data collection * Qualitative classification * Amplitude-domain distribution
* Data conversion * Stationarity * Time-domain distribution
* Data preprocessing * Periodicity * Spatial-domain distribution
* Nonstationarity (2-D and above)
* Frequency-domain distribution
§ 1.2 Data Engineering(5)
(b)Mathematical Model
Randomization Probability
design and test axioms
Uncertain Statistical Probability Math
Physical Experiment Space Representation
World
Statistical Probability
modeling modeling
Modeling and mapping
§ 1.2 Data Engineering(6)
(c)Engineering Model
Error
Algorithm +
Input Data Output
+
Array Information
Hardware
Additive error
Error can be additive or multiplicative.
§ 1.3 Scope(1)
1. Phases of Data Analysis
Data Acquisition
Signal Propagation
Source Path Pre-processing
Noise
Data Reduction
Generation Signal
Mechanism (S) Noise
(N)
*Reduced Signature
Analysis
* Features
Parameters
Data
Base Classification
S/N Max. (optimal)
Go/No Go Specific Problem
§ 1.3 Scope(2)
2.Information Estimation
(a) Hypothesis：The existence and uniqueness of true
information in data set.
(b) Objective：Maximize S/N under the constraints of
data, algorithm, and hardware.
Data Algorithm
S/N
Hardware
§ 1.3 Scope(3)
Ex. A Simple estimation
(a) Signal information
N
1
ˆ
Mean value: x
N
x
i 1
i
(b) Data set: xi
(c) Data constraint: N
(d) Noise
Error : || x ˆ x|| , “ || ||’’ : Norm
§ 1.3 Scope(4)
3.Main Information
(a) Single state:
Mean value: x
2 2
Mean square variance: x ( RMS x )
Probability density function: p ( x )
Autocorrelation function: R xx ( )
Power spectral density function: S xx ( )
§ 1.3 Scope(5)
(b) Multiple states (2-D for example)
Correlation coefficient: ( x, y )
Cross correlation: R xy ( )
Cross spectrum: S xy ( j )
2 S xy ( j )
Coherence function: r xy ( )
S xx ( ) S yy ( )
Joint probability density: p xy
§ 1.3 Scope(6)
4. System Applications
Parametric approach
Analysis
Non-parametric approach
System health
Operation
Engineering Quality control
system Transmission properties
Realization Source identification
Path identification
Test Performance testing
§ 1.4 Limitations of Course(1)
1. Content Limitation
Data engineering approach
Classical nonparametric analysis
Concern on stationary one-dimensional process
§ 1.4 Limitations of Course(2)
2. Knowledge Limitation
Decision
Intelligence
Engineering Knowledge
Signal Information
Random Data
