predictive analytics for coordinated optimization in...
TRANSCRIPT
NREL/PR-5D00-70018
August 28, 2017
Predictive Analytics for Coordinated Optimization in Distribution Systems
Rui Yang
Research Engineer
Power Systems Engineering Center
National Renewable Energy Laboratory
Workshop on Data Analytics for the Smart Grid Pullman, WA
2
• Increased Amount of Data in Power Systems
2 Motivation
3
• Data
o Nonpervasive
o Heterogeneous
o Highly variable
o Different resolution
3 Motivation
Topology data
Operations Optimization Control Planning
DER
Distribution Transmission
Load Load
4
• Data
o Nonpervasive
o Heterogeneous
o Highly variable
o Different resolution
4
Topology data
How?
How?
Real-Time Operations
How to use the data?
DER
Distribution Transmission
Motivation
Load Load
How to facilitate the real-time decision-making?
5
5 Power System Situational Awareness
Distribution Transmission
DER
Nodal Voltage
Load
Load
Monitor current states
6
6 Power System Situational Awareness
Distribution Transmission
DER
Future
Future
Nodal Voltage
Load
Load
Future
Monitor current states
Forecast future states
Future
7
• Renewable with Smart Inverters
o Able to adjust power generation
o Providing grid services
• Smart Loads
o Smart appliances
o Flexible power consumption
• Challenge – Lack of Coordination
o Not necessary to benefit the overall system operations
o Not fully utilizing the flexibility brought by these resources
Flexible Resources
Source: PV Magazine
Source: Microchip Technology Inc.
8
8 Predictive System Operations
Distribution Transmission
DER
Future
Future
Nodal Voltage
Load
Load
Future
Future
Coordinated Optimization
Set Points of Resources
9
• State Forecasting-Based Voltage Regulation [1, 2]
Applications
• Consumer Behavior-Aided Dispatch [3-5]
Coordinated Optimization
2 PM1 PM 3 PM 4 PM 5 PM 6 PM 7 PM 8 PM
0.89
0.9
0.91
0.92
0.93
0.94
Errors
Vo
lta
ge (
p.u
.)
Training Forecasting
Training data
Forecast dataTarget data
10
• Goals
o Accurately forecasting system states in the near future
o Prioritizing the control needs
• Approach
State Forecasting-Based Voltage Regulation
Dynamic Weights
Load Deviation
Power Set Point
Load Flexibility
PV Inverters
OPF
Loads
Available Power
State Forecasting
Forecasting OPF
11
Dynamically Weighted OPF
Power balance
Voltage constraints
PV plant
Smart load
Dynamically determined by the forecasted voltages
12
12
Voltage Magnitude
Voltage Angle
Results – State Forecasting Error
Accurate state forecast with machine learning methods
13
13
Voltage Violation
Voltage violations reduced significantly with state forecasting-based optimal scheduling
Results – Voltage Regulation
14
• Goals
o Actively engaging electricity consumers
o Achieving system-level control objectives without sacrificing consumers’ needs
• Integrated Optimization Approach
Consumer Behavior-Aided Dispatch
Load Deviation
Power Set Point
PV
Load Flexibility
DMS
HEMS
PV Inverters
OPF
Smart Loads
PV
Smart Loads
Substation
Available Power
Transmission
Distribution
15
Model-Based Load Forecasting
HEMS
User Preference
Weather Forecast
Model Predictive Control
Weights
Model
Load Forecast
Weather
16
12am 4am 8am 12pm 4pm 8pm 12am(+1)-5
0
5
Time
Tota
l Lo
ad (M
W)
opt
w/o c
12am 4am 8am 12pm 4pm 8pm 12am(+1)-0.2
0
0.2
0.4
0.6
0.8
1
Time
Vo
ltag
e V
iola
tio
n (p
.u.)
opt
w/o c
Total Load
Voltage Violation
Results – Distribution System Level
17
• Example – One House
12am 4am 8am 12pm 4pm 8pm 12am(+1)-15
-10
-5
0
5
10
15
Time
Load
(kW
)
ref
min
opt
max
Load Consumption
System performance improved without significant load deviation
Results – Home Level
18
• Predictive Analytics for Coordinated Optimization
o Data analytics methods to facilitate the decision-making
o Optimal coordination of various resources
• Ongoing Work
o Data-driven, model-based, and hybrid methods for resource and load forecasts [6]
o Integrated framework for system state estimation and forecasting [7]
o Incentives to drive desirable behaviors of consumers [8]
Summary
19
[1] Huaiguang Jiang and Yingchen Zhang, “Short-term distribution system state forecast based on optimal synchrophasor sensor placement and extreme learning machine,” IEEE PES General Meeting, Boston, MA, July 2016.
[2] Rui Yang, Huaiguang Jiang, and Yingchen Zhang, “Short-term state forecasting-based optimal voltage regulation in distribution systems,” IEEE Innovative Smart Grid Technologies, Arlington, VA, April 2017.
[3] Rui Yang and Yingchen Zhang, “Coordinated Optimization of Distributed Energy Resources and Smart Loads in Distribution Systems,” IEEE PES General Meeting, Boston, MA, July 2016.
[4] Rui Yang, Yingchen Zhang, Hongyu Wu, and Annabelle Pratt, “Coupling energy management systems at distribution and home levels,” IEEE Transactions on Smart Grid, under review.
[5] Yingchen Zhang, Rui Yang, Kaiqing Zhang, Huaiguang Jiang, and Jun Jason Zhang, “Consumption behavior analytics-aided energy forecasting and dispatch,” IEEE Intelligent Systems, vol. 32, no. 4, pp. 59-63, 2017.
[6] Huaiguang Jiang, Yingchen Zhang, Eduard Muljadi, Jun Jason Zhang, and Wenzhong Gao, “A short-term and high-resolution distribution system load forecasting approach using support vector regression with hybrid parameters optimization,” IEEE Transactions on Smart Grid.
[7] Yingchen Zhang, “Predictive Analytics for Energy Systems State Estimation,” panel presentation, IEEE PES General Meeting, Chicago, IL, July 2017.
[8] Rui Yang and Yingchen Zhang, “Three-phase AC optimal power flow based distribution locational marginal price,” IEEE Innovative Smart Grid Technologies, Arlington, VA, April 2017.
References
Collaborators:
Yingchen (YC) Zhang
Huaiguang Jiang
Andrey Bernstein
Contact:
Rui Yang
Email: [email protected]
Thank You!