D y ( S ) C DYNAMICAL SYSTEMS & CONTROL (DYSC) ir. Jasper Juchem, Prof. dr. ir. Clara Ionescu, dr. ir. Cosmin Copot CAN WE MAKE OUR LAMPS SMARTER? Distributed Model Predictive Control to reduce energy consumption Contact [email protected] www.ugent.be/ea/eemmecs/en/research/dysc Jasper Juchem @JasperJuchem 1. Problem Statement • Electrical power demand Climate agreements • Distribution demand 70% Office Buildings Solutions: A. Production B. Improve usage efficiency 2. Goal • Artificial light complementary and subordinate to sun light Variations in light intensity (e.g. clouds, flickering, etc.) > eye strain, discomfort, headache, etc. Solution: Model Predictive control (MPC) • Complexity ~ (# lamps) n (n>1) Solution: Distributed control = only interested in neighbours 3. Testbed • Real-time controller: dSPACE DS 1104 R&D board > Simulink® block diagrams Z1 Z2 Z3 Z6 Z4 Z5 Z7 Z8 D Disturbance N Neighbour r Reference LB Light bulb S Sensor VB Voltage buffer 4. Block diagram V cc DAC x LB x V cc 1kΩ LDR x ADC x Voltage Buffer Sensor MPC VB x LB x D 1x + + LB N I x S x S N + + + + D 2N D 2x r y 5. Hypothesis • Average insolation on a horizontal surface Ē [kWh/m²/day] > Data specific for Belgium ( eosweb.larc.nasa.gov/sse/) Maximal energy reduction: reduction 6. Analogy Population dynamics Urban Drainage system Lighting system Microgrids Population System Lighting environment Power dispatch Strategy Source reservoir Lighting zones Distributed generators Population mass Total inflow receptor res. Total available voltage Total demanded power Agent Flow unit Voltage unit Power unit Proportion of agents Proportion of flow Proportion of voltage Proportion of power Strategic distribution Flow distrib. in source res. Voltage split among LB Economic power dispatch Payoff of strategy Current volume Tracking error Marginal utility

ir. Jasper Juchem, Prof. dr. ir. Clara Ionescu, dr. ir

Download PDF Report

Upload
others
View
15
Download
0

Embed Size (px)

Citation preview

Page 1: ir. Jasper Juchem, Prof. dr. ir. Clara Ionescu, dr. ir

D y(S) CDYNAMICAL SYSTEMS & CONTROL (DYSC)ir. Jasper Juchem, Prof. dr. ir. Clara Ionescu, dr. ir. Cosmin Copot

CAN WE MAKE OUR LAMPS SMARTER?Distributed Model Predictive Control to reduce energy consumption

Contact

[email protected]/ea/eemmecs/en/research/dysc

Jasper Juchem

@JasperJuchem

1. Problem Statement • Electrical power demand Climate agreements • Distribution demand

70%

Office Buildings

Solutions: A. Production B. Improve usage efficiency

2. Goal • Artificial light complementary and subordinate to sun light Variations in light intensity (e.g. clouds, flickering, etc.) > eye strain, discomfort, headache, etc.

Solution: Model Predictive control (MPC) • Complexity ~ (# lamps)n (n>1)

Solution: Distributed control = only interested in neighbours

3. Testbed • Real-time controller: dSPACE DS 1104 R&D board > Simulink® block diagrams

Z1 Z2

Z7 Z8

D DisturbanceN Neighbourr Reference

LB Light bulbS SensorVB Voltage buffer

4. Block diagram

Vcc

DACx

LBx

Vcc

1kΩ

LDRx

ADCx

Voltage Buffer Sensor

MPC VBx LBx

D1x

LBN

SN++

D2N

D2x

r y

5. Hypothesis • Average insolation on a horizontal surface Ē [kWh/m²/day] > Data specific for Belgium (eosweb.larc.nasa.gov/sse/)

Maximal energy reduction: reduction

6. Analogy Population dynamics Urban Drainage system Lighting system Microgrids

Population System Lighting environment Power dispatch Strategy Source reservoir Lighting zones Distributed generators Population mass Total inflow receptor res. Total available voltage Total demanded power Agent Flow unit Voltage unit Power unit Proportion of agents Proportion of flow Proportion of voltage Proportion of power Strategic distribution Flow distrib. in source res. Voltage split among LB Economic power dispatch Payoff of strategy Current volume Tracking error Marginal utility