A Genetic Algorithm Approach to Optimize Dispatching for A Micro-grid Energy System with Renewable Energy Sources
Download as PPTX, PDF0 likes122 views
The document presents a genetic algorithm for optimizing dispatching in a microgrid energy system that incorporates renewable sources such as hydro, wind, and solar power. It details the objectives of reducing line loss and operating costs, while matching energy sources to load demands, and outlines a multi-step algorithm involving initialization, evolution through crossover and mutation, and final output of the fittest configuration. The study includes performance evaluations and network configurations that enhance energy resource utilization over a 24-hour period.
![Objectives:
• Network Optimization(Opening unnecessary line section)
and radial network reduces the line loss and removes the
circulating current throughout the network.
• Matching or opening available energy sources according to
load demand reduces the generation cost, operating cost
and saves the extra power.
• Most of the demand satisfy by the Hydro-electric power if
demand not exceed. [Ref]
[Ref]: 1 megawatt-hour of electricity costs $90.3 in 2011 to generate using hydropower and $144.30
to generate using solar collectors, according to the U.S. Energy Information](https://image.slidesharecdn.com/ga-191025204615/85/A-Genetic-Algorithm-Approach-to-Optimize-Dispatching-for-A-Micro-grid-Energy-System-with-Renewable-Energy-Sources-2-320.jpg)
![10/25/2019 3
Algorithm:
Step 1: Initialize population of size N
for each population Initialize chromosome of size M=24 [24 for hours from 1AM to 12 AM]
for every chromosome initialize two types of genes: [ 2 types for 2 objectives]
Initialize genes for sources of size 3(3 bit) [ for optimizing energy sources]
Initialize genes for fittest network() of size 14 [ for network optimization]
for every chromosome calculate chromosome fitness based on fittest from both genes
Step 2:
For G number of generation
Evolve N population through Crossover & Mutation
Step 3: Output the fittest chromosome
Procedure of fittest network():
Initialize population of power network of size P of Graph G=(V,E)
Initialize chromosome of size Q [ Here 14] by check()
For R number of generation evolve population through crossover and mutation
Output: Fittest network after r generation
end
Procedure of check():
1 Generate Q number of genes of size 1 ( Random number from 1 to 16) by randomly deleting two edges and
checking Strong Connectivity through remaining edges. [Applied DFS to check strongly connectivity]
If not Strongly connected, repeat 1
Otherwise output the network.
end](https://image.slidesharecdn.com/ga-191025204615/85/A-Genetic-Algorithm-Approach-to-Optimize-Dispatching-for-A-Micro-grid-Energy-System-with-Renewable-Energy-Sources-3-320.jpg)
![10/25/2019 10
Fitness functions:
𝑝𝑒𝑛𝑎𝑙𝑖𝑧𝑒𝑑 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛, 𝑓𝐷 =
1, 𝑖𝑓𝐷𝑡 < 𝑗=1
𝑛
𝑃 𝑇𝑗
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
If Hydro power 𝑃 𝑇1 > 𝐷𝑖 and 𝑃 𝑇2 = 0 and 𝑃 𝑇3 = 0 [To use Hydro-electric power most ]
𝑓𝐻 =0.001 [Award]
If Hydro power & Wind power 𝑃 𝑇1 + 𝑃 𝑇2 > 𝐷𝑖 and 𝑃 𝑇3 = 0 & Hydro power 𝑃 𝑇1 < 𝐷𝑖
𝑓𝐻𝑊 =0.001 [Award]
𝑖=1
𝑅
min 𝑓𝐿 ; R is the number of generation occurred for network reconfiguration
𝐿𝑜𝑠𝑠 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛, 𝑓𝐿 = 𝑃𝐿𝑜𝑠𝑠 = 𝑖∈𝑁𝑖 𝐼𝑗
2
𝑅𝑖
where 𝐼𝑗 =
𝑃 𝑗
𝑉
=
𝑃 𝑗
220𝑘𝑉
, 𝑗 = 1,2 𝑎𝑛𝑑 3 and 𝑁𝑖= number of nodes
𝑉min < 𝑉𝑗 < 𝑉m𝑎𝑥
Individual gene fitness 𝑓𝑖 = 𝑓𝐷 * 𝑓𝐻 * 𝑓𝐻𝑊 * 𝑓𝐿
Total chromosome fitness = 𝑖=1
24
𝐹𝑖](https://image.slidesharecdn.com/ga-191025204615/85/A-Genetic-Algorithm-Approach-to-Optimize-Dispatching-for-A-Micro-grid-Energy-System-with-Renewable-Energy-Sources-10-320.jpg)
A Genetic Algorithm Approach to Optimize Dispatching for A Micro-grid Energy System with Renewable Energy Sources
- 1. A Genetic Algorithm Approach to Optimize Dispatching for a Microgrid Energy System with Renewable Energy Sources by Sajib Sen, Kishor Datta Gupta, Subash Poudyal and Md Manjurul Ahsan
- 2. Objectives: • Network Optimization(Opening unnecessary line section) and radial network reduces the line loss and removes the circulating current throughout the network. • Matching or opening available energy sources according to load demand reduces the generation cost, operating cost and saves the extra power. • Most of the demand satisfy by the Hydro-electric power if demand not exceed. [Ref] [Ref]: 1 megawatt-hour of electricity costs $90.3 in 2011 to generate using hydropower and $144.30 to generate using solar collectors, according to the U.S. Energy Information
- 3. 10/25/2019 3 Algorithm: Step 1: Initialize population of size N for each population Initialize chromosome of size M=24 [24 for hours from 1AM to 12 AM] for every chromosome initialize two types of genes: [ 2 types for 2 objectives] Initialize genes for sources of size 3(3 bit) [ for optimizing energy sources] Initialize genes for fittest network() of size 14 [ for network optimization] for every chromosome calculate chromosome fitness based on fittest from both genes Step 2: For G number of generation Evolve N population through Crossover & Mutation Step 3: Output the fittest chromosome Procedure of fittest network(): Initialize population of power network of size P of Graph G=(V,E) Initialize chromosome of size Q [ Here 14] by check() For R number of generation evolve population through crossover and mutation Output: Fittest network after r generation end Procedure of check(): 1 Generate Q number of genes of size 1 ( Random number from 1 to 16) by randomly deleting two edges and checking Strong Connectivity through remaining edges. [Applied DFS to check strongly connectivity] If not Strongly connected, repeat 1 Otherwise output the network. end
- 4. 10/25/2019 4 Run Time: O(G(N(M(R(P(V+E)))))) V= nodes of load E= Line section between loads P= # of Population of power network g =(V+E) R=# of generation to evolve the power network g =(V+E) M= 24 # of genes having fittest power network and optimized energy sources from R generation N= # of M population G= # of generation to evolve N population Nested GA for two objective optimization GA for evolve fittest individual
- 5. Initial Chromosome: 7 2 4 5 2 7 4 6 7 1 0 3 7 2 4 5 2 1 4 6 1 4 6 7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 24 hours period 111 010 …. 111 … … … 111 Binary 3 bit representation of initial chromosome. As our system is using 3 power sources. Randomly selected number from 0 to 7 for 24 hours period 510/25/2019 Encoding Scheme:
- 6. Time (t) Power Demand (𝑫 𝒕) 6 1.5 MW Time (t) Hydro power 𝑃 𝑇1 Wind power 𝑃 𝑇2 Solar power 𝑃 𝑇3 Total power 6 220 kW 1934 kW 0 kW 2.154 MW For time t=6-7 am Demand, 𝐷𝑡=1.5 MW Generated power 𝑗=1 𝑛 𝑃 𝑇𝑗 = 220+1934+0= 2154 kW, where n=3 power sources 111 010 …. 111 … … … 111 t=6, gene= 111 610/25/2019 Encoding Scheme(Genes for Sources): An example
- 7. 4 1 2 3 5 6 7 8 9 12 11 13 1410 15 Hydro Power Wind Power Solar Power Figure : Red numbers are nodes, Blue numbers are line section 16 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Encoding Scheme(Genes for Fittest Network): Sample network 710/25/2019
- 8. Encoding Scheme(Gen es for Fittest Network): • Figure : Red numbers are nodes, Blue solid/dashed lines are line section active/opened, Blue numbers are line section number 10/25/2019 8
- 9. Crossover Operation between network configuration : P1: P2: O1: O2: 11 12 19 20 18 16 22 24 17 23 25 14 13 11 12 19 20 18 16 22 24 17 23 26 14 13 11 12 19 20 18 16 21 22 17 23 25 14 13 Sample Example: 910/25/2019
- 10. 10/25/2019 10 Fitness functions: 𝑝𝑒𝑛𝑎𝑙𝑖𝑧𝑒𝑑 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛, 𝑓𝐷 = 1, 𝑖𝑓𝐷𝑡 < 𝑗=1 𝑛 𝑃 𝑇𝑗 0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 If Hydro power 𝑃 𝑇1 > 𝐷𝑖 and 𝑃 𝑇2 = 0 and 𝑃 𝑇3 = 0 [To use Hydro-electric power most ] 𝑓𝐻 =0.001 [Award] If Hydro power & Wind power 𝑃 𝑇1 + 𝑃 𝑇2 > 𝐷𝑖 and 𝑃 𝑇3 = 0 & Hydro power 𝑃 𝑇1 < 𝐷𝑖 𝑓𝐻𝑊 =0.001 [Award] 𝑖=1 𝑅 min 𝑓𝐿 ; R is the number of generation occurred for network reconfiguration 𝐿𝑜𝑠𝑠 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛, 𝑓𝐿 = 𝑃𝐿𝑜𝑠𝑠 = 𝑖∈𝑁𝑖 𝐼𝑗 2 𝑅𝑖 where 𝐼𝑗 = 𝑃 𝑗 𝑉 = 𝑃 𝑗 220𝑘𝑉 , 𝑗 = 1,2 𝑎𝑛𝑑 3 and 𝑁𝑖= number of nodes 𝑉min < 𝑉𝑗 < 𝑉m𝑎𝑥 Individual gene fitness 𝑓𝑖 = 𝑓𝐷 * 𝑓𝐻 * 𝑓𝐻𝑊 * 𝑓𝐿 Total chromosome fitness = 𝑖=1 24 𝐹𝑖
- 11. Software used: Eclipse Library for performance measures and ploting: JFrame Application 10/25/2019 11
- 12. Source: https://www.eia.gov/ Typical Daily Load Demand : 1210/25/2019 Figure : Typical Daily Load Demand
- 13. Source: https://www.eia.gov/ Hourly Breakdown of Renewable Resources : 13 10/25/2019 Figure : Hourly Breakdown of Renewable Resources
- 16. 10/25/2019 16 Performance Measures(continued): Figure : Network Configured for 1AM – 2 AM period with optimized energy sources
- 17. 10/25/2019 17 Performance Measures(continued): Figure : Network Configured for 2AM – 3 AM period with optimized energy sources
- 18. 10/25/2019 18 Performance Measures(continued): Figure : Network Configured for 3AM – 4 AM period with optimized energy sources
- 19. 10/25/2019 19 Performance Measures(continued): Figure : Network Configured for 4AM – 5AM period with optimized energy sources
- 20. 10/25/2019 20 Performance Measures(continued): Figure : Network Configured for 5AM – 6AM period with optimized energy sources
- 21. 10/25/2019 21 Performance Measures(continued): Figure : Network Configured for 6AM – 7AM period with optimized energy sources
- 22. 10/25/2019 22 Performance Measures(continued): Figure : Network Configured for 7AM – 8AM period with optimized energy sources
- 23. 10/25/2019 23 Performance Measures(continued): Figure : Network Configured for 8AM – 9AM period with optimized energy sources
- 24. 10/25/2019 24 Performance Measures(continued): Figure : Network Configured for 9AM – 10AM period with optimized energy sources
- 25. 10/25/2019 25 Performance Measures(continued): Figure : Network Configured for 10AM – 11AM period with optimized energy sources
- 26. 10/25/2019 26 Performance Measures(continued): Figure : Network Configured for 11AM – 12AM period with optimized energy sources
- 27. 10/25/2019 27 Performance Measures(continued): Figure : Network Configured for 12PM – 1PM period with optimized energy sources
- 28. 10/25/2019 28 Performance Measures(continued): Figure : Network Configured for 1PM – 2PM period with optimized energy sources
- 29. 10/25/2019 29 Performance Measures(continued): Figure : Network Configured for 2PM – 3PM period with optimized energy sources
- 30. 10/25/2019 30 Performance Measures(continued): Figure : Network Configured for 3PM – 4PM period with optimized energy sources
- 31. 10/25/2019 31 Performance Measures(continued): Figure : Network Configured for 4PM – 5PM period with optimized energy sources
- 32. 10/25/2019 32 Performance Measures(continued): Figure : Network Configured for 5PM – 6PM period with optimized energy sources
- 33. 10/25/2019 33 Performance Measures(continued): Figure : Network Configured for 6PM – 7PM period with optimized energy sources
- 34. 10/25/2019 34 Performance Measures(continued): Figure : Network Configured for 7PM – 8PM period with optimized energy sources
- 35. 10/25/2019 35 Performance Measures(continued): Figure : Network Configured for 8PM – 9PM period with optimized energy sources
- 36. 10/25/2019 36 Performance Measures(continued): Figure : Network Configured for 9PM – 10PM period with optimized energy sources
- 37. 10/25/2019 37 Performance Measures(continued): Figure : Network Configured for 10PM – 11PM period with optimized energy sources
- 38. 10/25/2019 38 Performance Measures(continued): Figure : Network Configured for 11PM – 12PM period with optimized energy sources
- 39. 10/25/2019 39 Performance Measures(continued): Figure : Network Configured for 12PM – 1AM period with optimized energy sources