Preface (Second Edition)
Acknowledgments
Preface (First Edition)
A Third Need in Engineering Education
Intent
Approach and Scope
Assumed Background of Reader
Acknowledgements
Contents
1: Mathematical Models and Data Analysis
1.1 Forward and Inverse Approaches
1.1.1 Preamble
1.1.2 The Energy Problem and Importance of Buildings
1.1.3 Forward or Simulation Approach
1.1.4 Inverse or Data Analysis Approach
1.1.5 Discussion of Both Approaches
1.2 System Models
1.2.1 What Is a System Model?
1.2.2 Types of Models
1.3 Types of Data
1.3.1 Classification
1.3.2 Types of Uncertainty in Data
1.4 Mathematical Models
1.4.1 Basic Terminology
1.4.2 Block Diagrams
1.4.3 Mathematical Representation
1.4.4 Classification
1.4.5 Steady-State and Dynamic Models
1.5 Mathematical Modeling Approaches
1.5.1 Broad Categorization
1.5.2 Simulation or Forward Modeling
1.5.3 Inverse Modeling
1.5.4 Calibrated Simulation
1.6 Data Analytic Approaches
1.6.1 Data Mining or Knowledge Discovery
1.6.2 Machine Learning or Algorithmic Models
1.6.3 Introduction to Big Data
1.7 Data Analysis
1.7.1 Introduction
1.7.2 Basic Stages
1.7.3 Example of a Data Collection and Analysis System
1.8 Topics Covered in Book
Problems
References
2: Probability Concepts and Probability Distributions
2.1 Introduction
2.1.1 Classical Concept of Probability
2.1.2 Bayesian Viewpoint of Probability
2.1.3 Distinction Between Probability and Statistics
2.2 Classical Probability
2.2.1 Basic Terminology
2.2.2 Basic Set Theory Notation and Axioms of Probability
2.2.3 Axioms of Probability
2.2.4 Joint, Marginal, and Conditional Probabilities
2.2.5 Permutations and Combinations
2.3 Probability Distribution Functions
2.3.1 Density Functions
2.3.2 Expectations and Moments
2.3.3 Function of Random Variables
2.3.4 ChebyshevΒ΄s Theorem
2.4 Important Probability Distributions
2.4.1 Background
2.4.2 Distributions for Discrete Variables
2.4.3 Distributions for Continuous Variables
2.5 Bayesian Probability
2.5.1 BayesΒ΄ Theorem
2.5.2 Application to Discrete Probability Variables
2.5.3 Application to Continuous Probability Variables
2.6 Three Kinds of Probabilities
Problems
References
3: Data Collection and Preliminary Analysis
3.1 Sensors and Their Characteristics
3.2 Data Collection Systems
3.2.1 Generalized Measurement System
3.2.2 Types and Categories of Measurements
3.2.3 Data Recording Systems
3.3 Raw Data Validation and Preparation
3.3.1 Definitions
3.3.2 Limit Checks
3.3.3 Consistency Checks Involving Conservation Balances
3.3.4 Outlier Rejection by Visual Means
3.3.5 Handling Missing Data
3.4 Statistical Measures of Sample Data
3.4.1 Summary Descriptive Measures
3.4.2 Covariance and Pearson Correlation Coefficient
3.5 Exploratory Data Analysis (EDA)
3.5.1 What Is EDA?
3.5.2 Purpose of Data Visualization
3.5.3 Static Univariate Graphical Plots
3.5.4 Static Bi- and Multivariate Graphical Plots
3.5.5 Interactive and Dynamic Graphics
3.5.6 Basic Data Transformations
3.6 Overall Measurement Uncertainty
3.6.1 Need for Uncertainty Analysis
3.6.2 Basic Uncertainty Concepts: Random and Bias Errors
3.6.3 Random Uncertainty
3.6.4 Bias Uncertainty
3.6.5 Overall Uncertainty
3.6.6 ChauvenetΒ΄s Statistical Criterion of Data Rejection
3.7 Propagation of Errors
3.7.1 Taylor Series Method for Cross-Sectional Data
3.7.2 Monte Carlo Method for Error Propagation Problems
3.8 Planning a Non-Intrusive Field Experiment
Problems
References
4: Making Statistical Inferences from Samples
4.1 Introduction
4.2 Basic Univariate Inferential Statistics
4.2.1 Sampling Distribution and Confidence Interval of the Mean
4.2.2 Hypothesis Test for Single Sample Mean
4.2.3 Two Independent Sample and Paired Difference Tests on Means
4.2.4 Single and Two Sample Tests for Proportions
4.2.5 Single and Two Sample Tests of Variance
4.2.6 Tests for Distributions
4.2.7 Test on the Pearson Correlation Coefficient
4.3 ANOVA Test for Multi-Samples
4.3.1 Single-Factor ANOVA
4.3.2 TukeyΒ΄s Multiple Comparison Test
4.4 Tests of Significance of Multivariate Data
4.4.1 Introduction to Multivariate Methods
4.4.2 Hotteling T2 Test
4.5 Non-Parametric Tests
4.5.1 Signed and Rank Tests for Medians
4.5.2 Kruskal-Wallis Multiple Samples Test for Medians
4.5.3 Test on Spearman Rank Correlation Coefficient
4.6 Bayesian Inferences
4.6.1 Background
4.6.2 Estimating Population Parameter from a Sample
4.6.3 Hypothesis Testing
4.7 Some Considerations About Sampling
4.7.1 Random and Non-Random Sampling Methods
4.7.2 Desirable Properties of Estimators
4.7.3 Determining Sample Size During Random Surveys
4.7.4 Stratified Sampling for Variance Reduction
4.8 Resampling Methods
4.8.1 Basic Concept
4.8.2 Application to Probability Problems
4.8.3 Different Methods of Resampling
4.8.4 Application of Bootstrap to Statistical Inference Problems
4.8.5 Closing Remarks
Problems
References
5: Linear Regression Analysis Using Least Squares
5.1 Introduction
5.2 Regression Analysis
5.2.1 Objective of Regression Analysis
5.2.2 Ordinary Least Squares
5.3 Simple OLS Regression
5.3.1 Estimation of Model Parameters
5.3.2 Statistical Criteria for Model Evaluation
5.3.3 Inferences on Regression Coefficients and Model Significance
5.3.4 Model Prediction Uncertainty
5.4 Multiple OLS Regression
5.4.1 Higher Order Linear Models
5.4.2 Matrix Formulation
5.4.3 Point and Interval Estimation
5.4.4 Beta Coefficients and Elasticity
5.4.5 Partial Correlation Coefficients
5.4.6 Assuring Model Parsimony-Stepwise Regression
5.5 Applicability of OLS Parameter Estimation
5.5.1 Assumptions
5.5.2 Sources of Errors During Regression
5.6 Model Residual Analysis and Regularization
5.6.1 Detection of Ill-Conditioned Behavior
5.6.2 Leverage and Influence Data Points
5.6.3 Remedies for Nonuniform Residuals
5.6.4 Serially Correlated Residuals
5.6.5 Dealing with Misspecified Models
5.7 Other Useful OLS Regression Models
5.7.1 Zero-Intercept Models
5.7.2 Indicator Variables for Local Piecewise Models- Linear Splines
5.7.3 Indicator Variables for Categorical Regressor Models
5.8 Resampling Methods Applied to Regression
5.8.1 Basic Approach
5.8.2 Jackknife and k-Fold Cross-Validation
5.8.3 Bootstrap Method
5.9 Case Study Example: Effect of Refrigerant Additive on Chiller Performance
5.10 Parting Comments on Regression Analysis and OLS
Problems
References
6: Design of Physical and Simulation Experiments
6.1 Introduction
6.1.1 Types of Data Collection
6.1.2 Purpose of DOE
6.1.3 DOE Terminology
6.2 Overview of Different Statistical Methods
6.2.1 Different Types of ANOVA Tests
6.2.2 Link Between ANOVA and Regression
6.2.3 Recap of Basic Model Functional Forms
6.3 Basic Concepts
6.3.1 Levels, Discretization, and Experimental Combinations
6.3.2 Blocking
6.3.3 Unrestricted and Restricted Randomization
6.4 Factorial Designs
6.4.1 Full Factorial Design
6.4.2 2k Factorial Designs
6.4.3 Concept of Orthogonality
6.4.4 Fractional Factorial Designs
6.5 Block Designs
6.5.1 Complete Block Design
6.5.2 Latin Squares
6.6 Response Surface Designs
6.6.1 Applications
6.6.2 Methodology
6.6.3 First- and Second-Order Models
6.6.4 Central Composite Design and the Concept of Rotation
6.7 Simulation Experiments
6.7.1 Background
6.7.2 Similarities and Differences Between Physical and Simulation Experiments
6.7.3 Monte Carlo and Allied Sampling Methods
6.7.4 Sensitivity Analysis for Screening
6.7.5 Surrogate Modeling
6.7.6 Summary
Problems
References
7: Optimization Methods
7.1 Introduction
7.1.1 What Is Optimization?
7.1.2 Simple Example
7.2 Terminology and Classification
7.2.1 Definition of Terms
7.2.2 Categorization of Methods
7.2.3 Types of Objective Functions and Constraints
7.2.4 Sensitivity Analysis and Post-Optimality Analysis
7.3 Analytical Methods
7.3.1 Unconstrained Problems
7.3.2 Direct Substitution Method for Equality Constrained Problems
7.3.3 Lagrange Multiplier Method for Equality Constrained Problems
7.3.4 Problems with Inequality Constraints
7.3.5 Penalty Function Method
7.4 Numerical Unconstrained Search Methods
7.4.1 Univariate Methods
7.4.2 Multivariate Methods
7.5 Linear Programming (LP)
7.5.1 Standard Form
7.5.2 Example of a LP Problem
7.5.3 Linear Network Models
7.5.4 Example of Maximizing Flow in a Transportation Network
7.5.5 Mixed Integer Linear Programing (MILP)
7.5.6 Example of Reliability Analysis of a Power Network
7.6 Nonlinear Programming
7.6.1 Standard Form
7.6.2 Quadratic Programming
7.6.3 Popular Numerical Multivariate Search Algorithms
7.7 Illustrative Example: Integrated Energy System (IES) for a Campus
7.8 Introduction to Global Optimization
7.9 Examples of Dynamic Programming
Problems
References
8: Analysis of Time Series Data
8.1 Basic Concepts
8.1.1 Introduction
8.1.2 Terminology
8.1.3 Basic Behavior Patterns
8.1.4 Illustrative Data Set
8.2 General Model Formulations
8.3 Smoothing Methods
8.3.1 Arithmetic Moving Average (AMA)
8.3.2 Exponentially Weighted Moving Average (EWA)
8.3.3 Determining Structure by Cross-Validation
8.4 OLS Regression Models
8.4.1 Trend Modeling
8.4.2 Trend and Seasonal Models
8.4.3 Forecast Intervals
8.4.4 Fourier Series Models for Periodic Behavior
8.4.5 Interrupted Time Series
8.4.5.1 Abrupt One-Time Change in Time
8.4.5.2 Gradual Change Over Time
8.5 Stochastic Time Series Models
8.5.1 Introduction
8.5.2 ACF, PACF, and Data Detrending
8.5.2.1 Autocorrelation Function (ACF)
8.5.2.2 Partial Autocorrelation Function (PACF)
8.5.2.3 Detrending Data by Differencing
8.5.3 ARIMA Class of Models
8.5.3.1 Overview
8.5.3.2 ARMA Models
8.5.3.3 MA Models
8.5.3.4 AR Models
8.5.3.5 Identification and Forecasting
8.5.4 Recommendations on Model Identification
8.6 ARMAX or Transfer Function Models
8.6.1 Conceptual Approach and Benefit
8.6.2 Transfer Function Modeling of Linear Dynamic Systems
8.7 Quality Control and Process Monitoring Using Control Chart Methods
8.7.1 Background and Approach
8.7.2 Shewart Control Charts for Variables
8.7.2.1 Mean Charts
8.7.2.2 Range Charts
8.7.3 Shewart Control Charts for Attributes
8.7.4 Practical Implementation Issues of Control Charts
8.7.5 Time-Weighted Monitoring
8.7.5.1 Cusum Charts
8.7.5.2 EWMA Process
8.7.6 Concluding Remarks
Problems
References
9: Parametric and Non-Parametric Regression Methods
9.1 Introduction
9.2 Important Concepts in Parameter Estimation
9.2.1 Structural Identifiability
9.2.2 Ill-Conditioning
9.2.3 Numerical Identifiability
9.3 Dealing with Collinear Regressors: Variable Selection and Shrinkage
9.3.1 Problematic Issues
9.3.2 Principal Component Analysis and Regression
9.3.3 Ridge and Lasso Regression
9.3.4 Chiller Case Study Involving Collinear Regressors
9.3.5 Other Multivariate Methods
9.4 Going Beyond OLS
9.4.1 Background
9.4.2 Maximum Likelihood Estimation (MLE)
9.4.3 Generalized Linear Models (GLM)
9.4.4 Box-Cox Transformation
9.4.5 Logistic Functions
9.4.6 Error in Variables (EIV) and Corrected Least Squares
9.5 Non-Linear Parametric Regression
9.5.1 Detecting Non-Linear Correlation
9.5.2 Different Non-Linear Search Methods
9.5.3 Overview of Various Parametric Regression Methods
9.6 Non-Parametric Regression
9.6.1 Background
9.6.2 Extensions to Linear Models
9.6.3 Basis Functions
9.6.4 Polynomial Regression and Smoothing Splines
9.7 Local Regression: LOWESS Smoothing Method
9.8 Neural Networks: Multi-Layer Perceptron (MLP)
9.9 Robust Regression
Problems
References
Untitled
10: Inverse Methods for Mechanistic Models
10.1 Fundamental Concepts
10.1.1 Applicability
10.1.2 Approaches and Their Characteristics
10.1.3 Mechanistic Models
10.1.4 Scope of Chapter
10.2 Gray-Box Static Models
10.2.1 Basic Notions
10.2.2 Performance Models for Solar Photovoltaic Systems
10.2.2.1 Weather-Based Models
10.2.2.2 Weather and Cell Temperature-Based Models
10.2.3 Gray-Box and Black-Box Models for Water-Cooled Chillers
10.2.4 Sequential Stagewise Regression and Selection of Data Windows
10.2.5 Case Study of Non-Intrusive Sequential Parameter Estimation for Building Energy Flows
10.2.6 Application to Policy: Dose-Response
10.3 Certain Aspects of Data Collection
10.3.1 Types of Data Collection
10.3.2 Measures of Information Content
10.3.3 Functional Testing and Data Fusion
10.4 Gray-Box Models for Dynamic Systems
10.4.1 Introduction
10.4.2 Sequential Estimation of Thermal Network Model Parameters from Controlled Tests
10.4.3 Non-Intrusive Identification of Thermal Network Models and Parameters
10.4.4 State Space Representation and Compartmental Models
10.4.5 Example of a Compartmental Model
10.4.6 Practical Issues During Identification
10.5 Bayesian Regression and Parameter Estimation: Case Study
10.6 Calibration of Detailed Simulation Programs
10.6.1 Purpose
10.6.2 The Basic Issue
10.6.3 Detailed Simulation Models for Energy Use in Buildings
10.6.4 Uses of Calibrated Simulation
10.6.5 Causes of Differences
10.6.6 Definition of Terms
10.6.7 Raw Input Tuning (RIT)
10.6.8 Semi-Analytical Methods (SAM)
10.6.9 Physical Parameter Estimation (PPE)
10.6.10 Thoughts on Statistical Criteria for Goodness-of-Fit
Problems
References
11: Statistical Learning Through Data Analytics
11.1 Introduction
11.2 Distance as a Similarity Measure
11.3 Unsupervised Learning: Clustering Approaches
11.3.1 Types of Clustering Methods
11.3.2 Centroid-Based Partitional Clustering by K-Means
11.3.3 Density-Based Partitional Clustering Using DBSCAN
11.3.4 Agglomerative Hierarchical Clustering Methods
11.4 Supervised Learning: Statistical-Based Classification Approaches
11.4.1 Different Types of Approaches
11.4.2 Distance-Based Classification: k-Nearest Neighbors
11.4.3 Naive Bayesian Classification
11.4.4 Classical Regression-Based Classification
11.4.5 Discriminant Function Analysis
11.4.6 Neural Networks: Radial Basis Function (RBF)
11.4.7 Support Vector Machines (SVM)
11.5 Decision Tree-Based Classification Methods
11.5.1 Rule-Based Method and Decision-Tree Representation
11.5.2 Criteria for Tree Splitting
11.5.3 Classification and Regression Trees (CART)
11.5.4 Ensemble Method: Random Forest
11.6 Anomaly Detection Methods
11.6.1 Introduction
11.6.2 Graphical and Statistical Methods
11.6.3 Model-Based Methods
11.6.4 Data Mining Methods
11.7 Applications to Reducing Energy Use in Buildings
Problems
References
12: Decision-Making, Risk Analysis, and Sustainability Assessments
12.1 Introduction
12.1.1 Types of Decision-Making Problems and Applications
12.1.2 Purview of Reliability, Risk Analysis, and Decision-Making
12.1.3 Example of Discrete Decision-Making
12.1.4 Example of Chiller FDD
12.2 Single Criterion Decision-Making
12.2.1 General Framework
12.2.2 Representing Problem Structure: Influence Diagrams and Decision Trees
12.2.3 Single and Multi-Stage Decision Problems
12.2.4 Value of Perfect Information
12.2.5 Different Criteria for Outcome Evaluation
12.2.6 Discretizing Probability Distributions
12.2.7 Utility Value Functions for Modeling Risk Attitudes
12.2.8 Monte Carlo Simulation for First-Order and Nested Uncertainties
12.3 Risk Analysis
12.3.1 The Three Aspects
12.3.2 The Empirical Approach
12.3.3 Context of Environmental Risk to Humans
12.3.4 Other Areas of Application
12.4 Case Study: Risk Assessment of an Existing Building
12.5 Multi-Criteria Decision-Making (MCDM) Methods
12.5.1 Introduction and Description of Terms
12.5.2 Classification of Methods
12.5.3 Basic Mathematical Operations
12.6 Single Discipline MCDM Methods: Techno-Economic Analysis
12.6.1 Review
12.6.2 Consistent Attribute Scales
12.6.3 Inconsistent Attribute Scales: Dominance and Pareto Frontier
12.6.4 Case Study of Conflicting Criteria: Supervisory Control of an Engineered System
12.7 Sustainability Assessments: MCDM with Multi-Discipline Attribute Scales
12.7.1 Definitions and Scope
12.7.2 Indicators and Metrics
12.7.3 Sustainability Assessment Frameworks
12.7.4 Examples of Non, Semi-, and Fully-Aggregated Assessments
12.7.5 Two Case Studies: Structure-Based and Performance-Based
12.7.6 Closure
Problems
References
Appendices
Appendix: A: Statistical Tables (Tables A.1, A.2, A.3, A.4, A.5, A.6, A.7, A.8, A.9, A.10, A.11, A.12, and A.13)
Appendix B. Large Data Sets
Appendix C. Data Sets in Textbook
Appendix D: Solved Examples and Problems with Practical Relevance
Index