Introduction
Data analytics techniques for operation support
Applications of Data Analytics Techniques in Power Systems
Data Integration and Modeling
Data Quality and Validation
Summary and Conclusions
2. Content
• Introduction
• Data analytics techniques for operation support
• Applications of Data Analytics Techniques in Power Systems
• Data Integration and Modeling
• Data Quality and Validation
• Summary and Conclusions
4. Data Analytics for Power System
Operation Support
Assessment of the current and future state
of the system in order to be able to respond
with the correct measures to reach a
desired goal
• Contingency analysis
• Outage coordination
• Event detection and analysis
• Load and renewable forecast
• Models management –
component database
• Components rating calculation
• Compliance
• Special studies
Operator SituationalAwareness Processes in operation engineering
5. Situational Awareness Challenges
• Operators are becoming
supervisors of
automated processes
• Incorporates
“awareness” and
response from local
equipment
• Increase in alarms from
IEDs and sensors, PMU
data
6. Approach
What data is available?
Is it useful?
What data analytics techniques
and tools are available to obtain
actionable information from
data?
What technologies are
needed for handling data and
performing integration?
How quality of data can be
properly assessed and improved
to make the analytic solution
more valuable and reliable?
What is the present status of the use of data
analytics for power system operation support?
8. Data Analytics Technologies
Artificial intelligence (AI) is a branch of computer science
dedicated to create machines that reasoning, speech, and
vision like humans.
10. Artificial Intelligence
Speech Machine
Learning
Vision RoboticsExpert
Systems
Language
Processing
Planning
Spervised
Learning
Regression
SVM
KNN
Decision Trees
Naive-Bayes
Inductive Logic
Clustering
Association Rules
Data mining
Genetic Algorithms
Bayesian Networks
Neural Networks
Deep learning
Classification, Forecasting, Clustering, Optimization, Diagnosis, Prediction
Machine Learning Algorithms
y=f(x/)Model Estimation
Un Spervised
Learning
Pattern identification
Reinforcement
Learning
Reward
11. Applications of Machine Learning
Algorithms in Power Systems
Support vector machine
• Power system state estimation using
Smart Meters and PMUs in DG.
• Prediction of electricity demand in SG
• Forecasting: electricity demand, electricity
load, wind speed, power price, etc.
• Electrical networks faults diagnosis.
• Power system stability analysis.
• Identification of Root Causes in
Transmission Line Faults
[G. Santamaria ,et al.. Renewable Energy 85 (2016)]
Output:
+1: PowerSwing
−1:Fault
SVM classifier
Wind speed forecasting (WSF)
• Wind speed data from the Mexican Wind
Energy Technology Center (CERTE).
Power Swing detection.
• Voltage and current phasors, real and
reactive power measurements serve as
inputs of the model.
12. Deep learning (ANN)
• Fault location and classification
• Predicting and counteracting transient
instabilities
• Controlled islanding
• Under frequency load shedding
• Bad relay settings
• Load Forecasting
Applications of Machine Learning
Algorithms in Power Systems
13. Bayesian networks
• Power system state estimation using smart
meters and PMUs in DG.
• Prediction of electricity demand in SG
grids using Bayesian networks
• Power consumption (energy) and load
curves forecast
• Electrical networks faults diagnosis.
• Power system stability analysis
• Power system reliability studies
Bayesian networks of Events
• BN in which each node represents a
temporal event or change of state of a
variable.
• Detection of events; analysis of cause-
consequence of the events; and prediction
of future event and its time interval of
occurrence.
[G. Arroyo-Figueroa ,et al.. Appl. Intell. 23(2): 77-86, 2005]
Applications of Machine Learning
Algorithms in Power Systems
14. Task Area Algorithm (s)
Fault identification,
diagnosis
Power generation Bayesian networks, Association Rule
Transmission lines faults SVM, NN, BN, Decision Trees
Transformer Association Rule
Forecasting Wind power SVM, NN, BN
Solar power SVM, NN, BN, Regression (AR)
Load demand (DMS) SVM. NN, Regression
Power Price SVM
Estimation State estimation BN
Asset management Dynamic security Decision Trees
Condition Monitoring Regression
Alarm management Power generation Decision Trees, BN
Pattern recognition Events and disturbances Association Rule
Prediction Voltage collapse Regression
Analysis Power Growth Association Rule
Power system restoration, stability, reliability Association Rule, BN, NN
Impact of weather on demand Regression
Applications of Machine Learning
Algorithms in Power Systems
16. Use cases
• Operational decision support
• System situational awareness
• Renewable energy generation forecasting analytics
• Alarm processing and filtering
• Weather caused damage prediction
• Peak load management analytics
• Outage restoration analytics
• System oscillations detection (using PMU data)
• Real‐time voltage stability monitoring
• Fault location and root cause analysis
• Asset health assessment analytics
• Predictive asset maintenance analytics
• Power quality analytics
• Load research analytics
• Non-technical loss analytics
• Cyber security assessment analytics
17. Grid operation in power utilities becomes increasingly complex
21
Controlling and operating the power gridbecomes
increasingly complex as a result of:
Increased fluctuations due to renewable infeed.
Market driven system operation (FBMC, gridcodes)
New threats & risks (cyber/physical attacks,
natural hazards and severe weatherevents)
Aging infrastructure/workforce
Data analytics
Data visualisation
Situational awareness
Predictive analytics
Decision support
Will reduce complexity in operation
18. Data-driven system operation Moving from a
(reactive) control room to a (proactive) decision
support centre
22
ASSET
DATA
POWER
QUALITY
WEATHER
DATA
SMART
METER
DATA
CABLE
DATA
WORKFORCE
DATA
PMUDATA
GISDATA
Has
something
changed?
Do I need
to act?
Is this action
a good idea?
19. Complexity in system operation:
Solving the gap between data and actionable information
23
VisualAnalytics
21. A recent survey on application of digital technology
in Energy shows slow adoption rates
25
90%
digital technology is
crucial to the future
success of their utilities
70%
positive toward security
of storing data in the
cloud
Data Utilisation
>16%
estimated savings in
OPEX through
digitalization
15%
reduction of losses in
Transmission &
Distribution
Cost Savings
Key Findings
• Only 20% of utilities have
data analytics embedded in
their operational processes
• Only 23% of utilities have
reached a level of digital
maturity where they are
making capital expenditure
decisions based on
predictive analytics.
22. Data Integration and Modeling
• Data is only as good as the way it is packaged
• No advanced data management / analytics without
– accessible,
– flexible,
– scalable,
– comprehensive, and
– efficient data modeling
• Data modeling is about how to assemble data for secure and
reliable real-time grid operations
• In order to exchange the operational data between control
centers and throughout the industry, common data
exchange format and protocol need to be in place.
23. Information model and its usage
• The operating entities are
required to create and
maintain an accurate
model of their electric
systems.
• input data from various
sources , e.g.:
– generator owner,
– transmission owner,
– load,
– reliability coordinatorsExample from Dominion Virginia Power
EMS modelling data
24. Data Modeling
A need for common data model
• Evolutions of electrical grids induced by smart grids accelerate
changes in transmission and distribution:
– Volume of Data exchanges are increasing,
– Market deregulation has led to a proliferation of actors,
– Applications become more complex.
Actors of energy markets decided to use international standards.
• Growing need of smart grid stakeholders to deploy solutions
offering a semantic level of interoperability,
• Data modeling appears to be the key element and the foundation of
the smart grid framework.
• Furthermore, data modeling seems much more stable than
communication technologies
25. Data models for EPU
CIM (IEC 61 970, IEC 61 968, IEC 62 325) - power system management,
analysis, and related use cases (generation, market, and grid).
IEC 61850 - power utility automation use cases.
COSEM - metering and related use cases.
27. ISO 8000-8
• Framework for defining Data Quality
• Suitable mostly for organizations with well-defined
requirements
28. Data Quality Assessment
Approaches
• Bottom-up
– profiling tools and schema
inspections
– Generic, usage agnostic
– reveal indicators of potential areas
of data inconsistency
– Prone to false-positives
– Normally provides valuable input
to top-down method
• Top-down
– Domain experts and actual usage
scenarios to detect inconsistencies.
– Does not lend itself easily to
automation
Methods
Data quality
assessment
framework
Data
profiling
Data
interpolation
29. Data Quality Correction
Impact
assessment
Correction
and
cleaning
Scavenging
of essential
causes
Monitoring
and
Prevention
• Scope of influence
• Feasibility of correction
• Feasibility of prevention
• Control points in
• Data collection
• ETL
• Data analysis and application
• Cleaning in analysis and application
phase
• Finding the essential (root) causes
• Assessing them
• Eliminating them
• Establish monitoring
procedures for known DQ
issues
• Take action when identified
30. Conclusions
• To address the challenges and complexities of safe and reliable power grid
operation there is the need for a new generation of decision making tools
that combine scenario/contingency models with data analytics and
advanced visualization
• Data analytics can play a significant role in this process but there aresome
barriers for wide spread adoption in system operationsupport:
– Lack of understanding of the value and accuracy of the data analytics
technologies
– Standardized data structures
– Data quality
• Adoption rate of data analytics in system operation support is still lowbut
increasing, utilities have starts to realize the value and benefits
34