Test 2
Types of primary important (ACDs)
1) Explicit targets for system outputs are provided at every step. • 2) Explicit differentiable cost as a function of system variables is provided at every step. • 3) A graded cost is provided at each step but explicit relationship with system states is not given. 4) Ungraded reinforcement is provided when appropriate, e.g. binary outcome at the end of a game.
Multi-objective optimization/Multiple Criterion Decision Making (MCDM)
When an optimization problem involves more than one objective function, the task of finding one or more optimum solutions; f(X)=a1f1(X)+a2f2(X)
Single-objective optimization
When an optimization problem modeling a physical system involves only one objective function
Intelligence
Ability to comprehend, to understand and profit from experience, to interpret intelligence, having the capacity for thought and reason; creativity, skill, consciousness, emotion, and intuition
Smart Grid Data Analytics
Advanced Metering Infrastructure (AMI), Meter Data Management Systems (MDMS), Outage Management Systems (OMS), Distribution Management Systems (DMS), Enterprise Asset Management Systems (EAS)
Metadata
Any data that is used to describe other data
Electric Utility Challenges
Big data, communication, environment, sustainability, restoration, system scalability, system dynamic, business model change, IT architecture and services, cyber-security, workforce optimization
Distributed CSTMs
Co-existence of CSTMs (distributed) is essential for smart grid operations • Harmony • Coordination • Communication
High-Level Flow of Data in the Utility Network
Collection, Management, Analysis, Action
Cybersecurity analytics utility systems
Communications, advanced components, automated control systems, sensing and measurement, decision support, customer-facing systems
Stage 2 DA
Consider the possibilities of sampling and accurate information and re-evaluate the new EMV; Draw a new decision flow diagram; This should yield a best decision based on the highest EMV and/or the lowest expected loss
Typical optimization problems
DR management to maximize incentives, EMS, optimal power flow, forecasting models, Big Data
Human-in-the-Loop
Designing intuitive systems that users can operate with a minimum of cognitive friction is the goal of user-interface designers who realize the stakes are high
What is creating the perfect storm for grid modernization and smart electrification?
Economic drivers, carbon reduction, regulatory compliance, and an increase in the drive to provide residential, commercial, and industrial customer self-management of energy costs and consumption
Utilities' 3 primary domains for analytics
Enterprise analytics, grid operations analytics, and consumer analytics
Stage 1 DA
Evaluate the expected monetary value (EMV) from the profit and loss data, and their associated probabilities; Draw the decision flow tree which should yield a best decision based on the highest EMV and/or the lowest expected loss
ETL Data Analytics Infrastructure
Extract - Reading the data from a data source that could be in a variety of formats, including relational or raw data; Transform - Converting the extracted data from its current form into the form of the target database; Load - Writing the data into the target data warehouse
Constrained optimization problem
Find X = {x1, x2, ..., xn} which minimizes f(X) Where X is an n-dimensional vector called the design vector, f(X) is termed the objective function
Unconstrained optimization problem
Find X = {x1, x2, ..., xn} which minimizes f(X) Where X is an n-dimensional vector called the design vector, f(X) is termed the objective function
Asynchronous event messages
Grid devices with embedded processors generating messages under a variety of conditions, both as responses and commands
Dynamic programming
Highly nonlinear multivariate systems with many constraints represent some of the most difficult systems to control optimally (e.g. electric power grid, smart grid, robot manipulators); provides a computational technique to apply the principle of optimality to sequences of decisions which define an optimal control policy or trajectory
5 disciplines of CI
Immune systems, swarm intelligence, fuzzy systems, evolutionary computing, and neural networks
Situational Intelligence
Integrate historical and real-time data to implement near-future situational awareness
2 ways NNs remember the brain
Knowledge is acquired by the network from its environment through a learning process; Interneuron connection strengths, known as synaptic weights, are used to store acquired knowledge
Adaptive critic designs give brain-like intelligence for
MIMO system, nonlinear system, model uncertainties, random disturbance, learns over time, adaptive, robust
Benefits of cloud computing
More secure, focused cybersecurity, data privacy, flexible management for utilities, utilities can deploy data analytics applications for rapidly growing data volumes in a secure and scalable manner
Critical elements of data visualizations
Purpose, Choosing the right chart type, create effective views
Criterion, merit, or objective function
The criterion with respect to which the designs is optimized, when expressed as a function of the design variables
Challenges with decision support tools
The diversity in type and quality of information about a decision problem calls for methods and techniques that can assist in information processing
Why have utilities not been able to effectively use smart grid data?
The significant expertise deficit related to big data management, analytics, and data science; Data scientists not only need to know how to data wrangle, they must also know how to operate a variety of tools on a variety of platforms fed with vast amounts of varied data
Consumption data
This is the most often smart meter data, but any node that measures usage data may be included
Latency
Time it takes for a bit to travel from its sender to its receiver; very important in building a data analytic architecture
Computational Systems Thinking Machine (CSTM)
To handle an evolving, uncertain, variable and complex power system - three strands of thinking are needed for • Sense-making • Decision-making • Adaptation; In the center of all these strands exist a 'real-time wealth of knowledge' • Continuous refinement • Learns and unlearns
5 V's of Big Data
Volume, Variety, Velocity, Veracity, Variability
Swarm topology- ring (lbest)
the particles do not take into account all of the other particles but only their neighbors. This corresponds to a limited perception of the outside world
Adaptability principle (SI)
the population must be able to change behavior mode when it's worth the computational price
Proximity principle (SI)
the population should be able to carry out simple space and time computations
Quality principle (SI)
the population should be able to respond to quality factors in the environment
Stability principle (SI)
the population should not change its mode of behavior every time the environment changes
Diversity principle (SI)
the population should not commit its activities along excessively narrow channels
Swarm Intelligence
the property of a system whereby the collective behaviors of unsophisticated agents interacting locally with their environment cause coherent functional global patterns to emerge
Result of lack of IT-OT integration
uninformed or poor decision-making, lack of compliance, poor communications, inefficient field operations, inability to effectively report to external stakeholders
Utility Priorities for SG by Lockheed
visualization (presentation), situation awareness (decision making), predictive forecasting (prescriptive decision options)
What is DA based on?
what yields the greatest utility (or maximizes the occurrence of favorable outcomes) no matter how narrow the margin of improvement
Synchrophasor applications
Wide‐area situational awareness and intelligence, Outage restoration, load characterization, special protection schemes and islanding
Operation research
a branch of mathematics concerned with the application of scientific methods and techniques to decision making problems and with establishing the best or optimal solutions
Optimal control
a mathematical programming problem involving a number of stages, where each stage evolves from the preceding stage in a prescribed manner; defined by control variables and state variables
Analytic Hierarchy Process (AHP)
a multi-criteria decision-making approach; a decision support tool which can be used to solve complex decision problems; It uses a multi-level hierarchical structure of objectives, criteria, subcriteria, and alternatives
AHP is...
a nonlinear framework for carrying out both deductive and inductive thinking; without the use of the syllogism by considering several factors simultaneously, • allowing for dependence and feedback, and • making numerical tradeoffs to arrive at a synthesis or a conclusion
Brain intelligence
ability to cope with a large number of variables in parallel, in real time, in a noisy nonlinear and non-stationary environment
Globalized DHP (GDHP)
adapts a Critic network whose output is an approximation of J(R(t)), but adapts it so as to minimize errors in the implied derivatives of J (R(t)) (as well as J(R(t)) itself, with some weighting). GDHP tries to combine the best of HDP and DHP
Heuristic Dynamic Programming (HDP)
adapts a Critic network whose output is an approximation of J(R(t)). The temporal difference method of Barto, Sutton and Anderson turns out to be a special case of HDP
Dual Heuristic Programming (DHP)
adapts a Critic network whose outputs represent the derivatives of J(R(t))
Cloud computing
allows users to take benefit from all these technologies without the need for deep knowledge about or expertise with each one of them; virtualization that separates a physical computing device into one or more "virtual" devices, each of which can be easily used and managed to perform computing tasks
Big data
an all-encompassing term for any collection of data sets so large and complex that it becomes difficult to process using traditional data processing applications
Hadoop Data Analytics Infrastructure
an open-source framework that allows for the distributed processing of large data sets across clusters of computers; and the "Swiss Army knife of the 21st century"—that support the ability to process, manage, and give users the ability to directly consume data without moving it around; Scalable, cost-effective, flexible and fault-tolerant
Challenges of big data
analysis, capture, curation, search, sharing, storage, transfer, visualization, and privacy violations
Fuzzy systems
approximate reasoning with uncertain facts to infer new facts, with a degree of certainty associated with each fact; modeling common sense
Chart types for comparison and ranking
bar charts and pie charts; they encode quantitative values as length/size, making is easy to compare values
Artificial immune systems
biologically inspired models for immunization of engineering systems; to detect and eliminate non‐self materials, called antigens such as virus or cancer cells
Is AHP good for qualitative or quantitative analysis?
both
Chart types for distribution
boxplots, histograms
Wide-Area Monitoring and Visualization (WAMV) Systems
collect phasor data across an area as wide as an entire interconnection, which could be 100s of miles in size, in real-time and display it for operators to understand grid conditions
Define computational intellignce
computational models and tools of intelligence; capable of taking large raw numerical sensory data directly, processing them by exploiting the representational parallelism and pipelining the problem, generating reliable and timely responses, and withstanding high fault tolerance
Telemetry
continuous flow measurements of grid equipment parameters and other grid variables
4 applications of smart grid data analytics
customer satisfaction, reliability, operational efficiency, safety
Oscillographic
data made up of voltage and current waveform samples that can creat a graphical record
Analytics process flow
data processing, analytics, correlations and rules, visualization engine, visualization
Optimization techniques
decision system (DA/AHP), static (dynamic programming), adaptive dynamic programming (DHP/HDP), swarm and evolutionary computation (PSO/Heuristic programming), soft computing/Computational Intelligence (NNs/Fuzzy logic)
3 data analytic subsets
descriptive analytics, predictive analytics, and prescriptive analytics
Computing platforms
desktop computing, embedded processors, power PCs, GPUs, high processing computing, RTPIS lab
Adaptive critic designs
determines optimal control laws for a system by successively adapting two networks, namely an Action network (which dispenses control signals) and a Critic network (which learns the desired performance index for some function associated with the performance index).
Classification of Optimization Problems
existence of constraints, nature of the design variables, physical structure of the problem, nature of the equations involved, permissible values of the design variables, deterministic nature of the variables, separability of the functions, objective functions
Cellular Computational Networks
generally consists computational units connected to each other in an ordered distributed manner; suited to model complex systems with temporal and spatial dynamics
Complex data processing and analytics environment
hierarchical to distributed, multiple data classes, latencies
Prescriptive analytics
identify gaps in existing assets and establish sound asset management practices and programs (Mind-the-Gap)
Benefits of big data
improve the reliability and resiliency of electric grid, optimize the asset management and operations costs, share the data/intelligence for improved decision making, integrate legacy systems for improved data flow, improved data analytics and enterprise intelligence
List some ways to create effective views of data visualizations
limit number of colors and shapes in a single view, put most important data on the X or Y axis and less important data on color, size, or shape, avoid overloading your views
Chart types for trends over time
line charts, area charts, and bar charts with time on the x-axis and the measure on the y-axis
Operational Technology (OT)
manages the distribution operations, monitors infrastructure and control center-based systems, and oversees a lot of nonhuman inter- action between systems on the grid
Chart types for geographic data
maps
Neural networks
massively parallel distributed processor made up of simple processing units, which has the natural propensity for storing experiential knowledge and making it available for use
Models in analytics
models are the heart and longs of advanced analytics; it is a science and art to develop a model
Data analytics challenge
operations, energy trading, RT DR, asset management
Islanding and Restoration
operators can use synchrophasor data to bring equipment and load back into service without risking power instability or without experiencing unsuccessful reclosing attempts that prolong outages
AHP appications
planning, resource allocation, conflict resolution
What are critical for real-time monitoring?
predictions
Why is data visualization important?
provide an accessible way to see and understand trends, outliers, and patterns in data
Approximate Dynamic Programming
provide an approach to the optimal control and dynamic optimization of nonlinear systems whereby future rewards are estimated based upon state‐action transitions
Basic principles of SI
proximity principle, quality principle, diversity principle, stability principle, adaptability principle
Resilient grid
real-time grid data, predictive analytics, prescriptive analytics, operational response, planning models
Swarm topology- star (gbest)
results in the entire swarm particles trying to converge on one spot
Chart types for correlation
scatter plots, combine two line charts with a bar chart
Information Technology (IT)
staff typically manages the transactional side of the enterprise: billing, accounting, asset management, human resources, and customer records
3 types of learning methods for NNs
supervised, unsupervised, and reinforcement
Disturbance Detection and Alarming Studies
synchrophasor-based situational awareness can warn operators when the stress is increasing
Particle Swarm Optimization
system initially has a population of random solutions called particles; each particle has random velocity and memory that keeps track of previous best position and corresponding fitness
Classes of Utility data (Single source of truth)
telemetry, oscillographic, consumption data, asynchronous event messages, metadata
Optimization
the act of finding the best result under given circumstances; the process of finding the conditions that give the maximum or minimum value of a function
the curse of dimensionality
the computational burden caused by computational requirements being proportional to the number of discretization points which grow exponentially